---
title: Einen Orchestrierungsmodus erstellen
url: https://platform.claude.com/docs/de/build-with-claude/mid-conversation-effort-example
description: Erstelle einen Modus auf Sitzungsebene, der eine dauerhafte Zustimmung für Multiagenten-Fan-out erteilt und mit System-Nachrichten mitten im Gespräch ein- und ausgeschaltet wird.
---

Ein „orchestration mode“ (Orchestrierungsmodus) ist ein Schalter auf Sitzungsebene: Wenn er eingeschaltet ist, legt das Modell maximale Gründlichkeit hinter jede substanzielle Anfrage, erkundet die Aufgabe zunächst selbst und verteilt die Arbeit dann standardmäßig per Fan-out an parallele Subagenten. Wenn er ausgeschaltet ist, kehrt dasselbe Orchestrierungs-Tool zum Opt-in pro Anfrage zurück.

Der Modus ist kein API-Parameter. Er wird vollständig aus dokumentierten Bausteinen zusammengesetzt:

1. **Eine Effort-Stufe:** Anfragen laufen mit einem dokumentierten [Effort](https://platform.claude.com/docs/de/build-with-claude/effort)-Wert wie `xhigh`. Es gibt keine versteckte Stufe oberhalb der auf dieser Seite aufgeführten. Dieses Beispiel setzt Effort auf der obersten Ebene jeder Anfrage, was keinen Beta-Header erfordert.
2. **Eine Modus-Erinnerung:** Eine [„mid-conversation system message“ (System-Nachricht mitten im Gespräch)](https://platform.claude.com/docs/de/build-with-claude/mid-conversation-system-messages) teilt dem Modell mit, dass der Modus aktiv ist, mit einer einzeiligen Auffrischung alle paar Turns und einem Austrittshinweis, wenn der Modus ausgeschaltet wird. Das `system`-Feld auf oberster Ebene ändert sich nie, sodass das gecachte Präfix intakt bleibt.
3. **Dauerhafte Zustimmung in der Tool-Beschreibung:** Die Beschreibung des Orchestrierungs-Tools legt fest, dass das Modell, solange der Modus eingeschaltet ist, für jede substanzielle Aufgabe einen Workflow verfassen und ausführen soll, ohne vorher zu fragen.

<Note>
  Dieses Beispiel verwendet System-Nachrichten mitten im Gespräch; welche Modelle und Plattformen sie unterstützen, findest du unter [System-Nachrichten mitten im Gespräch](https://platform.claude.com/docs/de/build-with-claude/mid-conversation-system-messages). Der Fan-out selbst vervielfacht den Token-Verbrauch: Eine einzelne Anfrage kann viele Subagenten-Gespräche erzeugen, reserviere den Modus also für Arbeit, die die Kosten rechtfertigt.
</Note>

## Die Schleife einrichten

Das Beispiel besteht aus einer einzigen Datei. Die Konstanten steuern die Effort-Stufe, die Form des Fan-outs und wie oft die Modus-Auffrischung erneut gesendet wird. `MAX_CONCURRENT` begrenzt, wie viele Subagenten gleichzeitig laufen (die PHP-Portierung arbeitet sequenziell und ignoriert diesen Wert); `MAX_TOTAL_SUBTASKS` begrenzt, wie viele das Modell in einem einzelnen Workflow-Aufruf in die Warteschlange stellen darf. Die Trennung der beiden erlaubt es dem Modell, einen großen Rückstand zu planen, ohne alles auf einmal zu starten. Die `DOC_TEST_MODE`-Prüfung begrenzt die Schleifen auf einen einzigen Turn, wenn diese Umgebungsvariable gesetzt ist, sodass das automatisierte Docs-Harness validieren kann, dass die Datei kompiliert und schnell fertig wird, ohne die vollständige Orchestrierung auszuführen; lass sie ungesetzt, wenn du das Beispiel selbst ausführst.

<CodeGroup exclude="shell:cURL, shell:CLI">
  ```python Python
  import atexit
  import concurrent.futures
  import hashlib
  import json
  import os
  import shutil
  import subprocess
  import sys
  import tempfile
  import threading

  import anthropic

  client = anthropic.Anthropic()

  MODEL = "claude-opus-5-5"
  EFFORT = "xhigh"

  SYSTEM_PROMPT = (
      "You are a helpful general-purpose agent. Answer the user's request directly."
  )

  REQUEST_TIMEOUT_SECONDS = 600
  BASH_TIMEOUT_SECONDS = 60
  TOOL_RESULT_MAX_CHARS = 8000
  MAX_CONCURRENT = 10
  DOC_TEST_MODE = bool(os.environ.get("DOC_TEST_MODE"))
  MAX_TOTAL_SUBTASKS = 2 if DOC_TEST_MODE else 200
  MAX_SUBAGENT_TURNS = 1 if DOC_TEST_MODE else 15
  MAX_MAIN_TURNS = 1 if DOC_TEST_MODE else 30
  TURNS_BETWEEN_REFRESHERS = 10
  JOURNAL_PATH = os.environ.get("ORCH_JOURNAL") or "orchestration_journal.json"
  ```

  ```typescript TypeScript
  import { exec } from "node:child_process";
  import { createHash } from "node:crypto";
  import { mkdtempDisposable, readFile, rename, writeFile } from "node:fs/promises";
  import { tmpdir } from "node:os";
  import { join } from "node:path";

  import Anthropic from "@anthropic-ai/sdk";

  const client = new Anthropic();

  const MODEL = "claude-opus-5-5";
  const EFFORT = "xhigh";

  const SYSTEM_PROMPT =
    "You are a helpful general-purpose agent. Answer the user's request directly.";

  const REQUEST_TIMEOUT_SECONDS = 600;
  const BASH_TIMEOUT_SECONDS = 60;
  const TOOL_RESULT_MAX_CHARS = 8000;
  const MAX_CONCURRENT = 10;
  const DOC_TEST_MODE = Boolean(process.env.DOC_TEST_MODE);
  const MAX_TOTAL_SUBTASKS = DOC_TEST_MODE ? 2 : 200;
  const MAX_SUBAGENT_TURNS = DOC_TEST_MODE ? 1 : 15;
  const MAX_MAIN_TURNS = DOC_TEST_MODE ? 1 : 30;
  const TURNS_BETWEEN_REFRESHERS = 10;
  const JOURNAL_PATH = process.env.ORCH_JOURNAL || "orchestration_journal.json";
  ```

  ```csharp C#
  using System.Diagnostics;
  using System.Security.Cryptography;
  using System.Text;
  using System.Text.Json;
  using Anthropic;
  using Anthropic.Models.Messages;

  AnthropicClient client = new();

  const Model model = Model.ClaudeOpus5_5;
  var effort = Effort.Xhigh;

  const string systemPrompt = "You are a helpful general-purpose agent. Answer the user's request directly.";

  const int requestTimeoutSeconds = 600;
  const int bashTimeoutSeconds = 60;
  const int toolResultMaxChars = 8000;
  const int maxConcurrent = 10;
  var docTestMode = Environment.GetEnvironmentVariable("DOC_TEST_MODE") is { Length: > 0 };
  var maxTotalSubtasks = docTestMode ? 2 : 200;
  var maxSubagentTurns = docTestMode ? 1 : 15;
  var maxMainTurns = docTestMode ? 1 : 30;
  const int turnsBetweenRefreshers = 10;
  var journalPath = Environment.GetEnvironmentVariable("ORCH_JOURNAL") is { Length: > 0 } configuredPath
      ? configuredPath
      : "orchestration_journal.json";
  ```

  ```go Go
  import (
  	"bytes"
  	"cmp"
  	"context"
  	"crypto/sha256"
  	"encoding/hex"
  	"encoding/json"
  	"errors"
  	"fmt"
  	"log"
  	"os"
  	"os/exec"
  	"path/filepath"
  	"strings"
  	"sync"
  	"time"

  	"github.com/anthropics/anthropic-sdk-go"
  )

  var client = anthropic.NewClient()

  const (
  	modelID = anthropic.ModelClaudeOpus5_5
  	effort  = anthropic.OutputConfigEffortXhigh

  	systemPrompt = "You are a helpful general-purpose agent. Answer the user's request directly."

  	requestTimeoutSeconds  = 600
  	bashTimeoutSeconds     = 60
  	toolResultMaxChars     = 8000
  	maxConcurrent          = 10
  	turnsBetweenRefreshers = 10
  )

  var (
  	docTestMode      = os.Getenv("DOC_TEST_MODE") != ""
  	maxTotalSubtasks = ifTest(2, 200)
  	maxSubagentTurns = ifTest(1, 15)
  	maxMainTurns     = ifTest(1, 30)
  	journalPath      = cmp.Or(os.Getenv("ORCH_JOURNAL"), "orchestration_journal.json")
  )

  func ifTest(test, normal int) int {
  	if docTestMode {
  		return test
  	}
  	return normal
  }
  ```

  ```java Java
  // java.base (java.util, java.nio.file, java.util.concurrent, ...) wird in einer kompakten
  // Quelldatei implizit importiert, daher brauchen nur das SDK und Jackson Imports.
  import com.anthropic.client.AnthropicClient;
  import com.anthropic.client.okhttp.AnthropicOkHttpClient;
  import com.anthropic.core.JsonArray;
  import com.anthropic.core.JsonBoolean;
  import com.anthropic.core.JsonObject;
  import com.anthropic.core.JsonString;
  import com.anthropic.core.JsonValue;
  import com.anthropic.core.RequestOptions;
  import com.anthropic.helpers.MessageAccumulator;
  import com.anthropic.models.messages.ContentBlockParam;
  import com.anthropic.models.messages.Message;
  import com.anthropic.models.messages.MessageCreateParams;
  import com.anthropic.models.messages.MessageParam;
  import com.anthropic.models.messages.Model;
  import com.anthropic.models.messages.OutputConfig;
  import com.anthropic.models.messages.StopReason;
  import com.anthropic.models.messages.TextBlock;
  import com.anthropic.models.messages.Tool;
  import com.anthropic.models.messages.ToolBash20250124;
  import com.anthropic.models.messages.ToolResultBlockParam;
  import com.anthropic.models.messages.ToolUseBlock;
  import com.fasterxml.jackson.core.JsonProcessingException;
  import com.fasterxml.jackson.core.type.TypeReference;
  import com.fasterxml.jackson.databind.JsonNode;
  import com.fasterxml.jackson.databind.ObjectMapper;

  AnthropicClient client = AnthropicOkHttpClient.fromEnv();

  static final Model MODEL = Model.CLAUDE_OPUS_5_5;
  static final boolean DOC_TEST_MODE =
          !Objects.requireNonNullElse(System.getenv("DOC_TEST_MODE"), "").isEmpty();
  static final OutputConfig.Effort EFFORT = OutputConfig.Effort.XHIGH;

  static final String SYSTEM_PROMPT =
          "You are a helpful general-purpose agent. Answer the user's request directly.";

  static final int REQUEST_TIMEOUT_SECONDS = 600;
  static final RequestOptions REQUEST_OPTIONS =
          RequestOptions.builder().timeout(Duration.ofSeconds(REQUEST_TIMEOUT_SECONDS)).build();
  static final int BASH_TIMEOUT_SECONDS = 60;
  static final int TOOL_RESULT_MAX_CHARS = 8000;
  static final int MAX_CONCURRENT = 10;
  static final int MAX_TOTAL_SUBTASKS = DOC_TEST_MODE ? 2 : 200;
  static final int MAX_SUBAGENT_TURNS = DOC_TEST_MODE ? 1 : 15;
  static final int MAX_MAIN_TURNS = DOC_TEST_MODE ? 1 : 30;
  static final int TURNS_BETWEEN_REFRESHERS = 10;
  static final Path JOURNAL_PATH = Path.of(Optional.ofNullable(System.getenv("ORCH_JOURNAL"))
          .filter(path -> !path.isEmpty())
          .orElse("orchestration_journal.json"));
  ```

  ```php PHP
  use Anthropic\Client;
  use Anthropic\Messages\ToolUseBlock;

  $client = new Client();

  const MODEL = 'claude-opus-5-5';
  define('DOC_TEST_MODE', (string) getenv('DOC_TEST_MODE') !== '');
  const EFFORT = 'xhigh';

  const SYSTEM_PROMPT = 'You are a helpful general-purpose agent. Answer the user\'s request directly.';

  const REQUEST_TIMEOUT_SECONDS = 600;
  const BASH_TIMEOUT_SECONDS = 60;
  const TOOL_RESULT_MAX_CHARS = 8000;
  const MAX_CONCURRENT = 10;
  define('MAX_TOTAL_SUBTASKS', DOC_TEST_MODE ? 2 : 200);
  define('MAX_SUBAGENT_TURNS', DOC_TEST_MODE ? 1 : 15);
  define('MAX_MAIN_TURNS', DOC_TEST_MODE ? 1 : 30);
  const TURNS_BETWEEN_REFRESHERS = 10;
  define('JOURNAL_PATH', getenv('ORCH_JOURNAL') ?: 'orchestration_journal.json');
  ```

  ```ruby Ruby
  require "anthropic"
  require "digest"
  require "fileutils"
  require "json"
  require "open3"
  require "tmpdir"

  CLIENT = Anthropic::Client.new

  MODEL = "claude-opus-5-5"
  EFFORT = :xhigh

  SYSTEM_PROMPT = "You are a helpful general-purpose agent. Answer the user's request directly."

  REQUEST_TIMEOUT_SECONDS = 600
  BASH_TIMEOUT_SECONDS = 60
  TOOL_RESULT_MAX_CHARS = 8000
  MAX_CONCURRENT = 10
  DOC_TEST_MODE = !ENV["DOC_TEST_MODE"].to_s.empty?
  MAX_TOTAL_SUBTASKS = DOC_TEST_MODE ? 2 : 200
  MAX_SUBAGENT_TURNS = DOC_TEST_MODE ? 1 : 15
  MAX_MAIN_TURNS = DOC_TEST_MODE ? 1 : 30
  TURNS_BETWEEN_REFRESHERS = 10
  JOURNAL_PATH = ENV["ORCH_JOURNAL"].to_s.empty? ? "orchestration_journal.json" : ENV["ORCH_JOURNAL"]
  ```
</CodeGroup>

## Die Modus-Erinnerungen definieren

Die Erinnerungen sind absichtlich kurz. Sie schalten den Modus um und verweisen auf die Tool-Beschreibung, in der die schwergewichtigen Anweisungen stehen. Der vollständige Text wird einmal gesendet, wenn der Modus eingeschaltet wird, die Auffrischung wird erst nach mehreren Nutzer-Turns erneut gesendet, und der Austrittshinweis wird einmal gesendet, wenn der Modus ausgeschaltet wird.

<CodeGroup exclude="shell:cURL, shell:CLI">
  ```python Python
  MODE_ENTER = (
      "Orchestration mode is on: optimize for the most exhaustive, correct answer rather than "
      "the fastest one. Use the Workflow tool on every substantive task, sized to the problem's "
      "natural decomposition rather than the maximum the tool allows. See the Workflow tool's "
      "description for standing consent, granularity guidance, and quality patterns. Work solo "
      "only on conversational or trivial turns."
  )
  MODE_REFRESH = "Orchestration mode is still on. Use the Workflow tool; see its standing consent section."
  MODE_EXIT = (
      "Orchestration mode is off. The Workflow tool's standard opt-in rule applies again."
  )
  ```

  ```typescript TypeScript
  const MODE_ENTER =
    "Orchestration mode is on: optimize for the most exhaustive, correct answer rather than " +
    "the fastest one. Use the Workflow tool on every substantive task, sized to the problem's " +
    "natural decomposition rather than the maximum the tool allows. See the Workflow tool's " +
    "description for standing consent, granularity guidance, and quality patterns. Work solo " +
    "only on conversational or trivial turns.";
  const MODE_REFRESH =
    "Orchestration mode is still on. Use the Workflow tool; see its standing consent section.";
  const MODE_EXIT =
    "Orchestration mode is off. The Workflow tool's standard opt-in rule applies again.";
  ```

  ```csharp C#
  const string modeEnter =
      "Orchestration mode is on: optimize for the most exhaustive, correct answer rather than "
      + "the fastest one. Use the Workflow tool on every substantive task, sized to the problem's "
      + "natural decomposition rather than the maximum the tool allows. See the Workflow tool's "
      + "description for standing consent, granularity guidance, and quality patterns. Work solo "
      + "only on conversational or trivial turns.";
  const string modeRefresh =
      "Orchestration mode is still on. Use the Workflow tool; see its standing consent section.";
  const string modeExit =
      "Orchestration mode is off. The Workflow tool's standard opt-in rule applies again.";
  ```

  ```go Go
  const (
  	modeEnter = "Orchestration mode is on: optimize for the most exhaustive, correct answer rather than " +
  		"the fastest one. Use the Workflow tool on every substantive task, sized to the problem's " +
  		"natural decomposition rather than the maximum the tool allows. See the Workflow tool's " +
  		"description for standing consent, granularity guidance, and quality patterns. Work solo " +
  		"only on conversational or trivial turns."
  	modeRefresh = "Orchestration mode is still on. Use the Workflow tool; see its standing consent section."
  	modeExit    = "Orchestration mode is off. The Workflow tool's standard opt-in rule applies again."
  )
  ```

  ```java Java
  static final String MODE_ENTER = """
          Orchestration mode is on: optimize for the most exhaustive, correct answer rather than \
          the fastest one. Use the Workflow tool on every substantive task, sized to the problem's \
          natural decomposition rather than the maximum the tool allows. See the Workflow tool's \
          description for standing consent, granularity guidance, and quality patterns. Work solo \
          only on conversational or trivial turns.""";
  static final String MODE_REFRESH =
          "Orchestration mode is still on. Use the Workflow tool; see its standing consent section.";
  static final String MODE_EXIT =
          "Orchestration mode is off. The Workflow tool's standard opt-in rule applies again.";
  ```

  ```php PHP
  const MODE_ENTER =
      'Orchestration mode is on: optimize for the most exhaustive, correct answer rather than '
      . 'the fastest one. Use the Workflow tool on every substantive task, sized to the problem\'s '
      . 'natural decomposition rather than the maximum the tool allows. See the Workflow tool\'s '
      . 'description for standing consent, granularity guidance, and quality patterns. Work solo '
      . 'only on conversational or trivial turns.';
  const MODE_REFRESH =
      'Orchestration mode is still on. Use the Workflow tool; see its standing consent section.';
  const MODE_EXIT =
      'Orchestration mode is off. The Workflow tool\'s standard opt-in rule applies again.';
  ```

  ```ruby Ruby
  MODE_ENTER =
    "Orchestration mode is on: optimize for the most exhaustive, correct answer rather than " \
    "the fastest one. Use the Workflow tool on every substantive task, sized to the problem's " \
    "natural decomposition rather than the maximum the tool allows. See the Workflow tool's " \
    "description for standing consent, granularity guidance, and quality patterns. Work solo " \
    "only on conversational or trivial turns."
  MODE_REFRESH =
    "Orchestration mode is still on. Use the Workflow tool; see its standing consent section."
  MODE_EXIT =
    "Orchestration mode is off. The Workflow tool's standard opt-in rule applies again."
  ```
</CodeGroup>

## Dauerhafte Zustimmung in der Tool-Beschreibung erteilen

Das Workflow-Tool trägt den eigentlichen Verhaltensvertrag: die Opt-in-Regel, die dauerhafte Zustimmung, die gilt, solange der Modus eingeschaltet ist, Hinweise zur Granularität für die Dimensionierung des Fan-outs und die Qualitätsmuster, auf die das Modell zurückgreifen kann (eine Verifikationswelle, ein Vollständigkeitskritiker, mehrphasige Sequenzierung). Subagenten erhalten außerdem ein `report_findings`-Tool, damit ihre Ergebnisse als strukturiertes JSON statt als Fließtext zurückkommen, und das Bash-Tool ist das von Anthropic definierte `bash_20250124`-Tool, das lokal ausgeführt wird.

<CodeGroup exclude="shell:cURL, shell:CLI">
  ```python Python
  WORKFLOW_TOOL = {
      "name": "Workflow",
      "description": (
          "Orchestrate a multiagent workflow: split a large task into independent subtasks "
          "and run them as parallel agents, then collect their results.\n\n"
          "Opt-in: only use this tool when the user explicitly asks for a workflow, or when a "
          "system message confirms that orchestration mode is on.\n\n"
          "Quality patterns: adversarial verification (a second wave of agents checks the first "
          "wave's findings against the source), a completeness critic (one agent hunts for what "
          "the others missed), and multiphase sequencing (understand, design, implement, and "
          "review as separate workflow calls, reading results between phases). A useful default "
          "is hybrid: scout inline first to discover the work-list, then fan out over it.\n\n"
          "Granularity: scope each subtask to a distinct concern, component, or question rather "
          "than per line or per file section. Scale the count to what the user asked for: a "
          "focused review of a module of a few hundred lines rarely needs more than about ten "
          "subtasks; a broad audit of a large codebase can justify more.\n\n"
          "Standing consent: while a system message confirms orchestration mode is on, that "
          "opt-in is standing. Author and run a workflow for every substantive task by default, "
          "and lean toward verifying findings adversarially. Work solo only on conversational "
          "turns or trivial mechanical edits. When a system message says the mode is off, "
          "revert to the opt-in rule above."
      ),
      "input_schema": {
          "type": "object",
          "properties": {
              "subtasks": {
                  "type": "array",
                  "items": {"type": "string"},
                  "description": "Independent subtask prompts to run as parallel agents",
              }
          },
          "required": ["subtasks"],
      },
  }

  BASH_TOOL = {"type": "bash_20250124", "name": "bash"}

  REPORT_TOOL = {
      "name": "report_findings",
      "description": (
          "Report the final findings for your subtask. Call this exactly once, when you are "
          "done investigating; it ends your task."
      ),
      "input_schema": {
          "type": "object",
          "properties": {
              "summary": {
                  "type": "string",
                  "description": "Two or three sentences of synthesis",
              },
              "findings": {
                  "type": "array",
                  "items": {
                      "type": "object",
                      "properties": {
                          "claim": {
                              "type": "string",
                              "description": "The finding, one sentence",
                          },
                          "evidence": {
                              "type": "string",
                              "description": "How it was verified (file, line, or command output)",
                          },
                          "severity": {
                              "type": "string",
                              "enum": ["high", "medium", "low", "info"],
                          },
                      },
                      "required": ["claim", "evidence", "severity"],
                  },
              },
          },
          "required": ["summary", "findings"],
      },
  }
  ```

  ```typescript TypeScript
  const WORKFLOW_TOOL: Anthropic.Tool = {
    name: "Workflow",
    description: [
      "Orchestrate a multiagent workflow: split a large task into independent subtasks " +
        "and run them as parallel agents, then collect their results.",
      "Opt-in: only use this tool when the user explicitly asks for a workflow, or when a " +
        "system message confirms that orchestration mode is on.",
      "Quality patterns: adversarial verification (a second wave of agents checks the first " +
        "wave's findings against the source), a completeness critic (one agent hunts for what " +
        "the others missed), and multiphase sequencing (understand, design, implement, and " +
        "review as separate workflow calls, reading results between phases). A useful default " +
        "is hybrid: scout inline first to discover the work-list, then fan out over it.",
      "Granularity: scope each subtask to a distinct concern, component, or question rather " +
        "than per line or per file section. Scale the count to what the user asked for: a " +
        "focused review of a module of a few hundred lines rarely needs more than about ten " +
        "subtasks; a broad audit of a large codebase can justify more.",
      "Standing consent: while a system message confirms orchestration mode is on, that " +
        "opt-in is standing. Author and run a workflow for every substantive task by default, " +
        "and lean toward verifying findings adversarially. Work solo only on conversational " +
        "turns or trivial mechanical edits. When a system message says the mode is off, " +
        "revert to the opt-in rule above."
    ].join("\n\n"),
    input_schema: {
      type: "object",
      properties: {
        subtasks: {
          type: "array",
          items: { type: "string" },
          description: "Independent subtask prompts to run as parallel agents"
        }
      },
      required: ["subtasks"]
    }
  };

  const BASH_TOOL: Anthropic.ToolBash20250124 = { type: "bash_20250124", name: "bash" };

  const REPORT_TOOL: Anthropic.Tool = {
    name: "report_findings",
    description:
      "Report the final findings for your subtask. Call this exactly once, when you are " +
      "done investigating; it ends your task.",
    input_schema: {
      type: "object",
      properties: {
        summary: { type: "string", description: "Two or three sentences of synthesis" },
        findings: {
          type: "array",
          items: {
            type: "object",
            properties: {
              claim: { type: "string", description: "The finding, one sentence" },
              evidence: {
                type: "string",
                description: "How it was verified (file, line, or command output)"
              },
              severity: { type: "string", enum: ["high", "medium", "low", "info"] }
            },
            required: ["claim", "evidence", "severity"]
          }
        }
      },
      required: ["summary", "findings"]
    }
  };
  ```

  ```csharp C#
  Tool workflowTool = new()
  {
      Name = "Workflow",
      Description = """
          Orchestrate a multiagent workflow: split a large task into independent subtasks and run them as parallel agents, then collect their results.

          Opt-in: only use this tool when the user explicitly asks for a workflow, or when a system message confirms that orchestration mode is on.

          Quality patterns: adversarial verification (a second wave of agents checks the first wave's findings against the source), a completeness critic (one agent hunts for what the others missed), and multiphase sequencing (understand, design, implement, and review as separate workflow calls, reading results between phases). A useful default is hybrid: scout inline first to discover the work-list, then fan out over it.

          Granularity: scope each subtask to a distinct concern, component, or question rather than per line or per file section. Scale the count to what the user asked for: a focused review of a module of a few hundred lines rarely needs more than about ten subtasks; a broad audit of a large codebase can justify more.

          Standing consent: while a system message confirms orchestration mode is on, that opt-in is standing. Author and run a workflow for every substantive task by default, and lean toward verifying findings adversarially. Work solo only on conversational turns or trivial mechanical edits. When a system message says the mode is off, revert to the opt-in rule above.
          """,
      InputSchema = new()
      {
          Properties = new Dictionary<string, JsonElement>
          {
              ["subtasks"] = JsonSerializer.SerializeToElement(new
              {
                  type = "array",
                  items = new { type = "string" },
                  description = "Independent subtask prompts to run as parallel agents",
              }),
          },
          Required = ["subtasks"],
      },
  };

  ToolBash20250124 bashTool = new();

  Tool reportTool = new()
  {
      Name = "report_findings",
      Description =
          "Report the final findings for your subtask. Call this exactly once, when you are "
          + "done investigating; it ends your task.",
      InputSchema = new()
      {
          Properties = new Dictionary<string, JsonElement>
          {
              ["summary"] = JsonSerializer.SerializeToElement(new
              {
                  type = "string",
                  description = "Two or three sentences of synthesis",
              }),
              ["findings"] = JsonSerializer.SerializeToElement(new
              {
                  type = "array",
                  items = new
                  {
                      type = "object",
                      properties = new
                      {
                          claim = new { type = "string", description = "The finding, one sentence" },
                          evidence = new
                          {
                              type = "string",
                              description = "How it was verified (file, line, or command output)",
                          },
                          severity = new { type = "string", @enum = new[] { "high", "medium", "low", "info" } },
                      },
                      required = new[] { "claim", "evidence", "severity" },
                  },
              }),
          },
          Required = ["summary", "findings"],
      },
  };
  ```

  ```go Go
  var workflowTool = anthropic.ToolUnionParam{
  	OfTool: &anthropic.ToolParam{
  		Name: "Workflow",
  		Description: anthropic.String("Orchestrate a multiagent workflow: split a large task into independent subtasks " +
  			"and run them as parallel agents, then collect their results.\n\n" +
  			"Opt-in: only use this tool when the user explicitly asks for a workflow, or when a " +
  			"system message confirms that orchestration mode is on.\n\n" +
  			"Quality patterns: adversarial verification (a second wave of agents checks the first " +
  			"wave's findings against the source), a completeness critic (one agent hunts for what " +
  			"the others missed), and multiphase sequencing (understand, design, implement, and " +
  			"review as separate workflow calls, reading results between phases). A useful default " +
  			"is hybrid: scout inline first to discover the work-list, then fan out over it.\n\n" +
  			"Granularity: scope each subtask to a distinct concern, component, or question rather " +
  			"than per line or per file section. Scale the count to what the user asked for: a " +
  			"focused review of a module of a few hundred lines rarely needs more than about ten " +
  			"subtasks; a broad audit of a large codebase can justify more.\n\n" +
  			"Standing consent: while a system message confirms orchestration mode is on, that " +
  			"opt-in is standing. Author and run a workflow for every substantive task by default, " +
  			"and lean toward verifying findings adversarially. Work solo only on conversational " +
  			"turns or trivial mechanical edits. When a system message says the mode is off, " +
  			"revert to the opt-in rule above."),
  		InputSchema: anthropic.ToolInputSchemaParam{
  			Properties: map[string]any{
  				"subtasks": map[string]any{
  					"type":        "array",
  					"items":       map[string]any{"type": "string"},
  					"description": "Independent subtask prompts to run as parallel agents",
  				},
  			},
  			Required: []string{"subtasks"},
  		},
  	},
  }

  var bashTool = anthropic.ToolUnionParam{
  	OfBashTool20250124: &anthropic.ToolBash20250124Param{},
  }

  var reportTool = anthropic.ToolUnionParam{
  	OfTool: &anthropic.ToolParam{
  		Name: "report_findings",
  		Description: anthropic.String("Report the final findings for your subtask. Call this exactly once, when you are " +
  			"done investigating; it ends your task."),
  		InputSchema: anthropic.ToolInputSchemaParam{
  			Properties: map[string]any{
  				"summary": map[string]any{"type": "string", "description": "Two or three sentences of synthesis"},
  				"findings": map[string]any{
  					"type": "array",
  					"items": map[string]any{
  						"type": "object",
  						"properties": map[string]any{
  							"claim": map[string]any{"type": "string", "description": "The finding, one sentence"},
  							"evidence": map[string]any{
  								"type":        "string",
  								"description": "How it was verified (file, line, or command output)",
  							},
  							"severity": map[string]any{"type": "string", "enum": []string{"high", "medium", "low", "info"}},
  						},
  						"required": []string{"claim", "evidence", "severity"},
  					},
  				},
  			},
  			Required: []string{"summary", "findings"},
  		},
  	},
  }
  ```

  ```java Java
  static final Tool WORKFLOW_TOOL = Tool.builder()
          .name("Workflow")
          .description("""
                  Orchestrate a multiagent workflow: split a large task into independent subtasks \
                  and run them as parallel agents, then collect their results.

                  Opt-in: only use this tool when the user explicitly asks for a workflow, or when a \
                  system message confirms that orchestration mode is on.

                  Quality patterns: adversarial verification (a second wave of agents checks the first \
                  wave's findings against the source), a completeness critic (one agent hunts for what \
                  the others missed), and multiphase sequencing (understand, design, implement, and \
                  review as separate workflow calls, reading results between phases). A useful default \
                  is hybrid: scout inline first to discover the work-list, then fan out over it.

                  Granularity: scope each subtask to a distinct concern, component, or question rather \
                  than per line or per file section. Scale the count to what the user asked for: a \
                  focused review of a module of a few hundred lines rarely needs more than about ten \
                  subtasks; a broad audit of a large codebase can justify more.

                  Standing consent: while a system message confirms orchestration mode is on, that \
                  opt-in is standing. Author and run a workflow for every substantive task by default, \
                  and lean toward verifying findings adversarially. Work solo only on conversational \
                  turns or trivial mechanical edits. When a system message says the mode is off, \
                  revert to the opt-in rule above.""")
          .inputSchema(Tool.InputSchema.builder()
                  .properties(Tool.InputSchema.Properties.builder()
                          .putAdditionalProperty("subtasks", JsonValue.from(Map.of(
                                  "type", "array",
                                  "items", Map.of("type", "string"),
                                  "description", "Independent subtask prompts to run as parallel agents")))
                          .build())
                  .putAdditionalProperty("required", JsonValue.from(List.of("subtasks")))
                  .build())
          .build();

  static final ToolBash20250124 BASH_TOOL = ToolBash20250124.builder().build();

  static final Tool REPORT_TOOL = Tool.builder()
          .name("report_findings")
          .description("""
                  Report the final findings for your subtask. Call this exactly once, when you are \
                  done investigating; it ends your task.""")
          .inputSchema(Tool.InputSchema.builder()
                  .properties(Tool.InputSchema.Properties.builder()
                          .putAdditionalProperty("summary", JsonValue.from(Map.of(
                                  "type", "string",
                                  "description", "Two or three sentences of synthesis")))
                          .putAdditionalProperty("findings", JsonValue.from(Map.of(
                                  "type", "array",
                                  "items", Map.of(
                                          "type", "object",
                                          "properties", Map.of(
                                                  "claim", Map.of(
                                                          "type", "string",
                                                          "description", "The finding, one sentence"),
                                                  "evidence", Map.of(
                                                          "type", "string",
                                                          "description", "How it was verified (file, line, or command output)"),
                                                  "severity", Map.of(
                                                          "type", "string",
                                                          "enum", List.of("high", "medium", "low", "info"))),
                                          "required", List.of("claim", "evidence", "severity")))))
                          .build())
                  .putAdditionalProperty("required", JsonValue.from(List.of("summary", "findings")))
                  .build())
          .build();
  ```

  ```php PHP
  const WORKFLOW_TOOL = [
      'name' => 'Workflow',
      'description' =>
          'Orchestrate a multiagent workflow: split a large task into independent subtasks '
          . "and run them as parallel agents, then collect their results.\n\n"
          . 'Opt-in: only use this tool when the user explicitly asks for a workflow, or when a '
          . "system message confirms that orchestration mode is on.\n\n"
          . 'Quality patterns: adversarial verification (a second wave of agents checks the first '
          . 'wave\'s findings against the source), a completeness critic (one agent hunts for what '
          . 'the others missed), and multiphase sequencing (understand, design, implement, and '
          . 'review as separate workflow calls, reading results between phases). A useful default '
          . "is hybrid: scout inline first to discover the work-list, then fan out over it.\n\n"
          . 'Granularity: scope each subtask to a distinct concern, component, or question rather '
          . 'than per line or per file section. Scale the count to what the user asked for: a '
          . 'focused review of a module of a few hundred lines rarely needs more than about ten '
          . "subtasks; a broad audit of a large codebase can justify more.\n\n"
          . 'Standing consent: while a system message confirms orchestration mode is on, that '
          . 'opt-in is standing. Author and run a workflow for every substantive task by default, '
          . 'and lean toward verifying findings adversarially. Work solo only on conversational '
          . 'turns or trivial mechanical edits. When a system message says the mode is off, '
          . 'revert to the opt-in rule above.',
      'input_schema' => [
          'type' => 'object',
          'properties' => [
              'subtasks' => [
                  'type' => 'array',
                  'items' => ['type' => 'string'],
                  'description' => 'Independent subtask prompts to run as parallel agents',
              ],
          ],
          'required' => ['subtasks'],
      ],
  ];

  const BASH_TOOL = ['type' => 'bash_20250124', 'name' => 'bash'];

  const REPORT_TOOL = [
      'name' => 'report_findings',
      'description' =>
          'Report the final findings for your subtask. Call this exactly once, when you are '
          . 'done investigating; it ends your task.',
      'input_schema' => [
          'type' => 'object',
          'properties' => [
              'summary' => ['type' => 'string', 'description' => 'Two or three sentences of synthesis'],
              'findings' => [
                  'type' => 'array',
                  'items' => [
                      'type' => 'object',
                      'properties' => [
                          'claim' => ['type' => 'string', 'description' => 'The finding, one sentence'],
                          'evidence' => [
                              'type' => 'string',
                              'description' => 'How it was verified (file, line, or command output)',
                          ],
                          'severity' => ['type' => 'string', 'enum' => ['high', 'medium', 'low', 'info']],
                      ],
                      'required' => ['claim', 'evidence', 'severity'],
                  ],
              ],
          ],
          'required' => ['summary', 'findings'],
      ],
  ];
  ```

  ```ruby Ruby
  WORKFLOW_TOOL = {
    name: "Workflow",
    description:
      "Orchestrate a multiagent workflow: split a large task into independent subtasks " \
      "and run them as parallel agents, then collect their results.\n\n" \
      "Opt-in: only use this tool when the user explicitly asks for a workflow, or when a " \
      "system message confirms that orchestration mode is on.\n\n" \
      "Quality patterns: adversarial verification (a second wave of agents checks the first " \
      "wave's findings against the source), a completeness critic (one agent hunts for what " \
      "the others missed), and multiphase sequencing (understand, design, implement, and " \
      "review as separate workflow calls, reading results between phases). A useful default " \
      "is hybrid: scout inline first to discover the work-list, then fan out over it.\n\n" \
      "Granularity: scope each subtask to a distinct concern, component, or question rather " \
      "than per line or per file section. Scale the count to what the user asked for: a " \
      "focused review of a module of a few hundred lines rarely needs more than about ten " \
      "subtasks; a broad audit of a large codebase can justify more.\n\n" \
      "Standing consent: while a system message confirms orchestration mode is on, that " \
      "opt-in is standing. Author and run a workflow for every substantive task by default, " \
      "and lean toward verifying findings adversarially. Work solo only on conversational " \
      "turns or trivial mechanical edits. When a system message says the mode is off, " \
      "revert to the opt-in rule above.",
    input_schema: {
      type: "object",
      properties: {
        subtasks: {
          type: "array",
          items: {type: "string"},
          description: "Independent subtask prompts to run as parallel agents"
        }
      },
      required: ["subtasks"]
    }
  }.freeze

  BASH_TOOL = {type: "bash_20250124", name: "bash"}.freeze

  REPORT_TOOL = {
    name: "report_findings",
    description:
      "Report the final findings for your subtask. Call this exactly once, when you are " \
      "done investigating; it ends your task.",
    input_schema: {
      type: "object",
      properties: {
        summary: {type: "string", description: "Two or three sentences of synthesis"},
        findings: {
          type: "array",
          items: {
            type: "object",
            properties: {
              claim: {type: "string", description: "The finding, one sentence"},
              evidence: {
                type: "string",
                description: "How it was verified (file, line, or command output)"
              },
              severity: {type: "string", enum: ["high", "medium", "low", "info"]}
            },
            required: ["claim", "evidence", "severity"]
          }
        }
      },
      required: ["summary", "findings"]
    }
  }.freeze
  ```
</CodeGroup>

## Das Bash-Tool lokal ausführen

Der Bash-Handler führt den angeforderten Befehl mit einem Timeout aus, erfasst stdout und stderr kombiniert und kürzt das Ergebnis, damit ein außer Kontrolle geratener Befehl das Kontextfenster nicht überfluten kann. Befehle laufen in dem Verzeichnis, aus dem du das Beispiel startest; um es auf ein Projekt zu richten, musst du es also dort starten. Wenn `DOC_TEST_MODE` gesetzt ist, gibt das Harness Bash stattdessen ein kleines Wegwerf-Fixture-Verzeichnis, das beim Beenden entfernt wird. Es gibt hier keine Sandbox: Der Befehl läuft mit den Berechtigungen des Prozesses, der das Beispiel gestartet hat. Der Übersichtlichkeit halber führt dieses Beispiel jeden Aufruf in einer frischen Subshell aus, statt die persistente Sitzung aufrechtzuerhalten, die der `bash_20250124`-Vertrag beschreibt; ein Produktionsagent sollte das Tool mit einer langlebigen Shell hinterlegen, damit Arbeitsverzeichnis, Umgebung und die `restart`-Aktion sich wie dokumentiert verhalten.

<CodeGroup exclude="shell:cURL, shell:CLI">
  ```python Python
  # Führe bash dort aus, wo das Beispiel gestartet wurde. Im DOC_TEST_MODE verweist der Docs-Harness
  # stattdessen auf ein Wegwerf-Fixture-Verzeichnis, das beim Beenden entfernt wird.
  if DOC_TEST_MODE:
      WORK_DIR = tempfile.mkdtemp(prefix="orchestration-")
      atexit.register(shutil.rmtree, WORK_DIR, ignore_errors=True)
      with open(os.path.join(WORK_DIR, "sample.py"), "w") as fixture:
          fixture.write(
              "def fib(n):\n"
              "    return n if n < 2 else fib(n - 1) + fib(n - 2)\n\n"
              "print(fib(10))\n"
          )
  else:
      WORK_DIR = os.getcwd()


  def run_bash(command: str) -> tuple[str, bool]:
      """Run a shell command and return (output, is_error). No sandbox: example code only."""
      print(f"[bash] {command}", file=sys.stderr)
      try:
          proc = subprocess.run(
              ["bash", "-c", command],
              cwd=WORK_DIR,
              capture_output=True,
              text=True,
              errors="replace",
              timeout=BASH_TIMEOUT_SECONDS,
          )
      except subprocess.TimeoutExpired:
          return f"command timed out after {BASH_TIMEOUT_SECONDS}s", True
      output = (proc.stdout + proc.stderr).strip() or "(no output)"
      if len(output) > TOOL_RESULT_MAX_CHARS:
          output = (
              output[:TOOL_RESULT_MAX_CHARS]
              + f"\n(truncated at {TOOL_RESULT_MAX_CHARS} chars)"
          )
      if proc.returncode != 0:
          output = f"(exit code {proc.returncode})\n{output}"
      return output, proc.returncode != 0


  def handle_bash_block(block) -> tuple[str, bool]:
      if block.input.get("restart") is True:
          return "Shell restarted.", False
      command = block.input.get("command")
      if not isinstance(command, str) or not command:
          return "bash error: no command was provided.", True
      return run_bash(command)
  ```

  ```typescript TypeScript
  type ToolOutcome = { output: string; isError: boolean };

  // Führt bash dort aus, wo das Beispiel gestartet wurde. Im DOC_TEST_MODE verweist der Docs-Harness
  // stattdessen ein temporäres Fixture-Verzeichnis, das beim Programmende entfernt wird.
  await using fixtureDir = DOC_TEST_MODE
    ? await mkdtempDisposable(join(tmpdir(), "orchestration-"))
    : null;
  const WORK_DIR = fixtureDir?.path ?? process.cwd();
  if (fixtureDir) {
    await writeFile(
      join(WORK_DIR, "sample.py"),
      `def fib(n):
      return n if n < 2 else fib(n - 1) + fib(n - 2)

  print(fib(10))
  `
    );
  }

  // Führt einen Shell-Befehl aus und gibt dessen Ausgabe zurück. Keine Sandbox: nur Beispielcode.
  function runBash(command: string): Promise<ToolOutcome> {
    console.error(`[bash] ${command}`);
    const options = {
      shell: "/bin/bash",
      cwd: WORK_DIR,
      timeout: BASH_TIMEOUT_SECONDS * 1000,
      maxBuffer: 16 * 1024 * 1024
    };
    return new Promise((resolve) => {
      exec(command, options, (error, stdout, stderr) => {
        if (error?.killed) {
          resolve({ output: `command timed out after ${BASH_TIMEOUT_SECONDS}s`, isError: true });
          return;
        }
        const exitCode = error === null ? 0 : typeof error.code === "number" ? error.code : 1;
        let output = (stdout + stderr).trim() || "(no output)";
        const codePoints = [...output];
        if (codePoints.length > TOOL_RESULT_MAX_CHARS) {
          const kept = codePoints.slice(0, TOOL_RESULT_MAX_CHARS).join("");
          output = `${kept}\n(truncated at ${TOOL_RESULT_MAX_CHARS} chars)`;
        }
        if (exitCode !== 0) {
          output = `(exit code ${exitCode})\n${output}`;
        }
        resolve({ output, isError: exitCode !== 0 });
      });
    });
  }

  async function handleBashBlock(block: Anthropic.ToolUseBlock): Promise<ToolOutcome> {
    const { command, restart } = block.input as { command?: string; restart?: boolean };
    if (restart === true) {
      return { output: "Shell restarted.", isError: false };
    }
    if (!command) {
      return { output: "bash error: no command was provided.", isError: true };
    }
    return runBash(command);
  }
  ```

  ```csharp C#
  // Führe bash dort aus, wo das Beispiel gestartet wurde. Im DOC_TEST_MODE verweist das
  // Docs-Harness stattdessen auf ein Wegwerf-Fixture-Verzeichnis, das beim Beenden entfernt wird.
  var workDir = Environment.CurrentDirectory;
  if (docTestMode)
  {
      workDir = Directory.CreateTempSubdirectory("orchestration-").FullName;
      await File.WriteAllTextAsync(Path.Combine(workDir, "sample.py"), """
          def fib(n):
              return n if n < 2 else fib(n - 1) + fib(n - 2)

          print(fib(10))

          """);
      var fixtureDir = workDir;
      AppDomain.CurrentDomain.ProcessExit += (_, _) =>
      {
          try { Directory.Delete(fixtureDir, recursive: true); }
          catch { /* Best-effort cleanup; the OS tmp sweeper handles leftovers. */ }
      };
  }

  // Führe einen Shell-Befehl aus und gib Ausgabe plus Fehler-Flag zurück. Keine Sandbox: nur Beispielcode.
  async Task<(string Output, bool IsError)> RunBash(string command)
  {
      Console.Error.WriteLine($"[bash] {command}");
      using var process = Process.Start(new ProcessStartInfo("bash")
      {
          ArgumentList = { "-c", command },
          WorkingDirectory = workDir,
          RedirectStandardOutput = true,
          RedirectStandardError = true,
      });
      if (process is null)
      {
          return ("bash error: the shell process failed to start.", true);
      }
      var stdoutTask = process.StandardOutput.ReadToEndAsync();
      var stderrTask = process.StandardError.ReadToEndAsync();
      using var timeout = new CancellationTokenSource(TimeSpan.FromSeconds(bashTimeoutSeconds));
      try
      {
          await process.WaitForExitAsync(timeout.Token);
      }
      catch (OperationCanceledException)
      {
          process.Kill(entireProcessTree: true);
          // Lass die Reader-Tasks abschließen, bevor der Prozess disposed wird.
          try
          {
              await Task.WhenAll(stdoutTask, stderrTask);
          }
          catch
          {
              // Die Ausgabe wird bei Timeout verworfen, also werden Reader-Fehler ebenfalls ignoriert.
          }
          return ($"command timed out after {bashTimeoutSeconds}s", true);
      }
      var output = (await stdoutTask + await stderrTask).Trim();
      if (output.Length == 0)
      {
          output = "(no output)";
      }
      if (output.Length > toolResultMaxChars)
      {
          output = output[..toolResultMaxChars] + $"\n(truncated at {toolResultMaxChars} chars)";
      }
      if (process.ExitCode != 0)
      {
          output = $"(exit code {process.ExitCode})\n{output}";
      }
      return (output, process.ExitCode != 0);
  }

  // Führe einen vom Modell angeforderten bash-Tool-Aufruf aus.
  async Task<(string Output, bool IsError)> HandleBashBlock(ToolUseBlock block)
  {
      if (block.Input.GetValueOrDefault("restart").ValueKind is JsonValueKind.True)
      {
          return ("Shell restarted.", false);
      }
      return block.Input.GetValueOrDefault("command") is { ValueKind: JsonValueKind.String } command
          && command.GetString() is { Length: > 0 } commandText
          ? await RunBash(commandText)
          : ("bash error: no command was provided.", true);
  }
  ```

  ```go Go
  // Führe bash dort aus, wo das Beispiel gestartet wurde. In DOC_TEST_MODE verweist das
  // Docs-Harness stattdessen auf ein Wegwerf-Fixture-Verzeichnis, das beim Beenden entfernt wird.
  var workDir = func() string {
  	if !docTestMode {
  		dir, err := os.Getwd()
  		if err != nil {
  			log.Fatal(err)
  		}
  		return dir
  	}
  	dir, err := os.MkdirTemp("", "orchestration-")
  	if err != nil {
  		log.Fatal(err)
  	}
  	const fixture = `def fib(n):
      return n if n < 2 else fib(n - 1) + fib(n - 2)

  print(fib(10))
  `
  	if err := os.WriteFile(filepath.Join(dir, "sample.py"), []byte(fixture), 0o644); err != nil {
  		log.Fatal(err)
  	}
  	return dir
  }()

  // runBash führt einen Shell-Befehl aus und gibt dessen Ausgabe plus ein Fehler-Flag zurück.
  // Keine Sandbox: nur Beispielcode.
  func runBash(ctx context.Context, command string) (string, bool) {
  	fmt.Fprintf(os.Stderr, "[bash] %s\n", command)
  	ctx, cancel := context.WithTimeout(ctx, bashTimeoutSeconds*time.Second)
  	defer cancel()
  	cmd := exec.CommandContext(ctx, "bash", "-c", command)
  	cmd.Dir = workDir
  	combined, err := cmd.CombinedOutput()
  	if errors.Is(ctx.Err(), context.DeadlineExceeded) {
  		return fmt.Sprintf("command timed out after %ds", bashTimeoutSeconds), true
  	}
  	output := strings.TrimSpace(string(combined))
  	if output == "" {
  		output = "(no output)"
  	}
  	if runes := []rune(output); len(runes) > toolResultMaxChars {
  		output = string(runes[:toolResultMaxChars]) + fmt.Sprintf("\n(truncated at %d chars)", toolResultMaxChars)
  	}
  	if err == nil {
  		return output, false
  	}
  	if exitErr, ok := errors.AsType[*exec.ExitError](err); ok {
  		return fmt.Sprintf("(exit code %d)\n%s", exitErr.ExitCode(), output), true
  	}
  	return fmt.Sprintf("(%s)\n%s", err, output), true
  }

  // handleBashBlock führt einen vom Modell angeforderten bash-Tool-Aufruf aus.
  func handleBashBlock(ctx context.Context, block anthropic.ToolUseBlock) (string, bool) {
  	var input struct {
  		Command string `json:"command"`
  		Restart bool   `json:"restart"`
  	}
  	if err := json.Unmarshal(block.Input, &input); err != nil {
  		return fmt.Sprintf("bash error: could not parse input: %s", err), true
  	}
  	if input.Restart {
  		return "Shell restarted.", false
  	}
  	if input.Command == "" {
  		return "bash error: no command was provided.", true
  	}
  	return runBash(ctx, input.Command)
  }
  ```

  ```java Java
  record ToolOutput(String output, boolean isError) {}

  // Führe bash dort aus, wo das Beispiel gestartet wurde. Im DOC_TEST_MODE verweist das Docs-Harness
  // stattdessen auf ein Wegwerf-Fixture-Verzeichnis, das beim Beenden entfernt wird.
  static final Path WORK_DIR = createWorkDir();

  static Path createWorkDir() {
      if (!DOC_TEST_MODE) {
          return Path.of(System.getProperty("user.dir"));
      }
      try {
          var dir = Files.createTempDirectory("orchestration-");
          Files.writeString(dir.resolve("sample.py"), """
                  def fib(n):
                      return n if n < 2 else fib(n - 1) + fib(n - 2)

                  print(fib(10))
                  """);
          Runtime.getRuntime().addShutdownHook(new Thread(() -> {
              // Best-Effort-Aufräumen; Reste entfernt der tmp-Cleaner des Betriebssystems.
              try (var paths = Files.walk(dir)) {
                  paths.sorted(Comparator.reverseOrder()).forEach(path -> {
                      try { Files.deleteIfExists(path); } catch (IOException _) {}
                  });
              } catch (IOException _) {}
          }));
          return dir;
      } catch (IOException error) {
          throw new UncheckedIOException(error);
      }
  }

  // Führe einen Shell-Befehl aus und gib Ausgabe plus Fehler-Flag zurück. Keine Sandbox: nur Beispielcode.
  ToolOutput runBash(String command) throws InterruptedException {
      System.err.println("[bash] " + command);
      Process process;
      try {
          process = new ProcessBuilder("bash", "-c", command)
                  .directory(WORK_DIR.toFile())
                  .redirectErrorStream(true)
                  .start();
      } catch (IOException error) {
          return new ToolOutput("(" + error + ")", true);
      }
      // Leere stdout in einem anderen Thread, damit eine volle Pipe das Timeout-Warten unten nicht blockiert.
      CompletableFuture<String> outputReader = CompletableFuture.supplyAsync(() -> {
          try (var stdout = process.getInputStream()) {
              return new String(stdout.readAllBytes(), StandardCharsets.UTF_8);
          } catch (IOException _) {
              return "";
          }
      });
      if (!process.waitFor(BASH_TIMEOUT_SECONDS, TimeUnit.SECONDS)) {
          process.destroyForcibly();
          outputReader.cancel(true);
          return new ToolOutput("command timed out after " + BASH_TIMEOUT_SECONDS + "s", true);
      }
      String output = outputReader.join().strip();
      if (output.isEmpty()) {
          output = "(no output)";
      } else if (output.length() > TOOL_RESULT_MAX_CHARS) {
          output = output.substring(0, TOOL_RESULT_MAX_CHARS)
                  + "\n(truncated at " + TOOL_RESULT_MAX_CHARS + " chars)";
      }
      int exitCode = process.exitValue();
      return exitCode == 0
              ? new ToolOutput(output, false)
              : new ToolOutput("(exit code " + exitCode + ")\n" + output, true);
  }

  // Die Eingabefelder eines Tool-Aufrufs oder eine leere Map, falls das Modell kein Objekt gesendet hat.
  Map<String, JsonValue> toolInput(ToolUseBlock toolUse) {
      return toolUse._input() instanceof JsonObject object ? object.values() : Map.of();
  }

  // Führe einen vom Modell angeforderten bash-Tool-Aufruf aus.
  ToolOutput handleBashBlock(ToolUseBlock toolUse) throws InterruptedException {
      Map<String, JsonValue> input = toolInput(toolUse);
      if (input.get("restart") instanceof JsonBoolean restart && restart.value()) {
          return new ToolOutput("Shell restarted.", false);
      }
      String command = input.get("command") instanceof JsonString text ? text.value() : "";
      if (command.isEmpty()) {
          return new ToolOutput("bash error: no command was provided.", true);
      }
      return runBash(command);
  }
  ```

  ```php PHP
  // Führe bash dort aus, wo das Beispiel gestartet wurde. Im DOC_TEST_MODE verweist der Docs-Harness
  // stattdessen auf ein Wegwerf-Fixture-Verzeichnis, das beim Beenden entfernt wird.
  if (DOC_TEST_MODE) {
      $workDir = sys_get_temp_dir() . '/orchestration-' . bin2hex(random_bytes(8));
      if (!mkdir($workDir, 0700)) {
          throw new \RuntimeException("could not create working directory {$workDir}");
      }
      file_put_contents(
          $workDir . '/sample.py',
          "def fib(n):\n"
          . "    return n if n < 2 else fib(n - 1) + fib(n - 2)\n\n"
          . "print(fib(10))\n",
      );
      register_shutdown_function(function () use ($workDir): void {
          foreach (glob($workDir . '/*') ?: [] as $entry) {
              @unlink($entry);
          }
          @rmdir($workDir);
      });
  } else {
      $workDir = getcwd() ?: '.';
  }
  define('WORK_DIR', $workDir);

  /**
   * Run a shell command and return [output, isError]. The coreutils timeout command
   * enforces the time limit. No sandbox: example code only.
   */
  function runBash(string $command): array
  {
      fwrite(STDERR, "[bash] {$command}\n");
      // Benötigt GNU coreutils 'timeout'. Auf macOS: brew install coreutils, oder durch gtimeout ersetzen.
      exec(
          'cd ' . escapeshellarg(WORK_DIR) . ' && timeout ' . BASH_TIMEOUT_SECONDS
              . ' bash -c ' . escapeshellarg($command) . ' 2>&1',
          $outputLines,
          $exitCode,
      );
      if ($exitCode === 124) {
          return ['command timed out after ' . BASH_TIMEOUT_SECONDS . 's', true];
      }
      $output = trim(implode("\n", $outputLines));
      if ($output === '') {
          $output = '(no output)';
      }
      if (mb_strlen($output) > TOOL_RESULT_MAX_CHARS) {
          $output = mb_substr($output, 0, TOOL_RESULT_MAX_CHARS)
              . "\n(truncated at " . TOOL_RESULT_MAX_CHARS . ' chars)';
      }
      if ($exitCode !== 0) {
          $output = "(exit code {$exitCode})\n{$output}";
      }
      return [$output, $exitCode !== 0];
  }

  /** Execute one bash tool call requested by the model. */
  function handleBashBlock(ToolUseBlock $block): array
  {
      if (($block->input['restart'] ?? null) === true) {
          return ['Shell restarted.', false];
      }
      $command = $block->input['command'] ?? '';
      if (!is_string($command) || $command === '') {
          return ['bash error: no command was provided.', true];
      }
      return runBash($command);
  }
  ```

  ```ruby Ruby
  # Führe bash dort aus, wo das Beispiel gestartet wurde. In DOC_TEST_MODE lenkt der Docs-Harness
  # es stattdessen auf ein Wegwerf-Fixture-Verzeichnis um, das beim Beenden entfernt wird.
  WORK_DIR =
    if DOC_TEST_MODE
      Dir.mktmpdir("orchestration-").tap do |dir|
        File.write(File.join(dir, "sample.py"), <<~PYTHON)
          def fib(n):
              return n if n < 2 else fib(n - 1) + fib(n - 2)

          print(fib(10))
        PYTHON
        at_exit { FileUtils.remove_entry(dir, true) }
      end
    else
      Dir.pwd
    end

  # Tool-Input kommt als Hash oder als roher JSON-String vom Streaming-
  # Akkumulator an; normalisiere beide Formen zu einem Hash mit String-Keys.
  def parse_tool_input(raw)
    return raw.transform_keys(&:to_s) if raw.is_a?(Hash)
    parsed = JSON.parse(raw.to_s) rescue nil
    parsed.is_a?(Hash) ? parsed : {}
  end

  # Führe einen Shell-Befehl aus und gib [output, is_error] zurück. Keine Sandbox: nur Beispielcode.
  def run_bash(command)
    warn "[bash] #{command}"
    begin
      stdin, stdout_and_stderr, wait_thr = Open3.popen2e("bash", "-c", command, pgroup: true, chdir: WORK_DIR)
      stdin.close
      reader = Thread.new { stdout_and_stderr.read.scrub }
      # Erzwinge das Zeitlimit mit einer Deadline auf Basis der monotonen Uhr, damit ein Befehl mit Timeout
      # beendet wird, statt im Hintergrund weiterzulaufen.
      deadline = Process.clock_gettime(Process::CLOCK_MONOTONIC) + BASH_TIMEOUT_SECONDS
      until wait_thr.join(0.1)
        next if Process.clock_gettime(Process::CLOCK_MONOTONIC) < deadline

        begin
          Process.kill("-TERM", wait_thr.pid)
        rescue Errno::ESRCH
        end
        unless wait_thr.join(2)
          begin
            Process.kill("-KILL", wait_thr.pid)
          rescue Errno::ESRCH
          end
        end
        wait_thr.join(5)
        reader.join(1) || reader.kill
        stdout_and_stderr.close rescue nil
        return ["command timed out after #{BASH_TIMEOUT_SECONDS}s", true]
      end
      status = wait_thr.value
      output = reader.value.strip
      stdout_and_stderr.close
      output = "(no output)" if output.empty?
      if output.length > TOOL_RESULT_MAX_CHARS
        output = "#{output[0, TOOL_RESULT_MAX_CHARS]}\n(truncated at #{TOOL_RESULT_MAX_CHARS} chars)"
      end
      output = "(exit code #{status.exitstatus})\n#{output}" unless status.success?
      [output, !status.success?]
    rescue Errno::ENOENT => e
      return ["bash error: #{e.message}", true]
    end
  end

  # Führe einen vom Modell angeforderten bash-Tool-Aufruf aus.
  def handle_bash_block(block)
    input = parse_tool_input(block.input)
    return ["Shell restarted.", false] if input["restart"] == true

    command = input["command"]
    return ["bash error: no command was provided.", true] unless command.is_a?(String) && !command.empty?

    run_bash(command)
  end

  # Wandle Response-Inhalte in Request-förmige Parameter um. Der Streaming-Akkumulator
  # liefert tool_use-Input als rohen JSON-String und enthält Felder, die nur in der Response vorkommen,
  # forme daher jeden Block ins Request-Schema um, bevor er zurückgegeben wird.
  def assistant_content_param(content)
    content.map do |block|
      case block
      when Anthropic::Models::ToolUseBlock
        input = parse_tool_input(block.input)
        {type: "tool_use", id: block.id, name: block.name, input: input}
      when Anthropic::Models::TextBlock
        {type: "text", text: block.text}
      when Anthropic::Models::ThinkingBlock
        {type: "thinking", thinking: block.thinking, signature: block.signature}
      when Anthropic::Models::RedactedThinkingBlock then {type: "redacted_thinking", data: block.data}
      else
        block.to_h
      end
    end
  end
  ```
</CodeGroup>

## Einen Subagenten ausführen

Jede Workflow-Teilaufgabe wird zu einer eigenen kleinen Agentenschleife mit dem Bash-Tool, die mit demselben Effort wie die Hauptschleife läuft. Ein Timeout pro Anfrage begrenzt jeden API-Aufruf, sodass eine abgebrochene Verbindung nur einen Subagenten beeinträchtigt, statt den gesamten Lauf zu blockieren.

<CodeGroup exclude="shell:cURL, shell:CLI">
  ```python Python
  def run_subagent(model: str, prompt: str) -> str:
      """One subagent: a small nested agent loop with the bash tool plus report_findings.
      Subagents inherit the main loop's effort level."""
      subagent_system = (
          "You are one agent in a larger parallel fan-out, assigned a single subtask. "
          "Investigate it directly, using bash to check facts rather than guessing, and finish "
          "by calling report_findings exactly once. Return findings, not narration."
      )
      messages = [{"role": "user", "content": prompt}]
      for _ in range(MAX_SUBAGENT_TURNS):
          with client.messages.stream(
              model=model,
              max_tokens=64000,
              system=subagent_system,
              output_config={"effort": EFFORT},
              tools=[BASH_TOOL, REPORT_TOOL],
              messages=messages,
              timeout=REQUEST_TIMEOUT_SECONDS,
          ) as stream:
              response = stream.get_final_message()
          messages.append({"role": "assistant", "content": response.content})
          if response.stop_reason == "pause_turn":
              continue
          if response.stop_reason != "tool_use":
              text = "".join(
                  block.text for block in response.content if block.type == "text"
              )
              if response.stop_reason == "max_tokens":
                  text += "\n\n(warning: subagent response was truncated at max_tokens)"
              return text
          tool_results = []
          report = None
          for block in response.content:
              if block.type != "tool_use":
                  continue
              match block.name:
                  case "report_findings":
                      report = json.dumps(block.input, indent=2)
                      output, is_error = "Findings recorded.", False
                  case "bash":
                      output, is_error = handle_bash_block(block)
                  case _:
                      output, is_error = f"unknown tool: {block.name}", True
              tool_results.append(
                  {
                      "type": "tool_result",
                      "tool_use_id": block.id,
                      "content": output,
                      "is_error": is_error,
                  }
              )
          if report is not None:
              return report
          messages.append({"role": "user", "content": tool_results})
      return "(subagent hit the turn limit before finishing)"
  ```

  ```typescript TypeScript
  // Ein Subagent: eine kleine verschachtelte Agenten-Schleife mit dem bash-Tool plus report_findings.
  // Subagenten erben das Effort-Level der Hauptschleife.
  async function runSubagent(model: string, prompt: string): Promise<string> {
    const subagentSystem =
      "You are one agent in a larger parallel fan-out, assigned a single subtask. " +
      "Investigate it directly, using bash to check facts rather than guessing, and finish " +
      "by calling report_findings exactly once. Return findings, not narration.";
    const messages: Anthropic.MessageParam[] = [{ role: "user", content: prompt }];
    for (let turn = 0; turn < MAX_SUBAGENT_TURNS; turn++) {
      const response = await client.messages
        .stream(
          {
            model,
            max_tokens: 64000,
            system: subagentSystem,
            output_config: { effort: EFFORT },
            tools: [BASH_TOOL, REPORT_TOOL],
            messages
          },
          { signal: AbortSignal.timeout(REQUEST_TIMEOUT_SECONDS * 1000) }
        )
        .finalMessage();
      messages.push({ role: "assistant", content: response.content });
      if (response.stop_reason === "pause_turn") {
        continue;
      }
      if (response.stop_reason !== "tool_use") {
        let text = response.content
          .flatMap((block) => (block.type === "text" ? [block.text] : []))
          .join("");
        if (response.stop_reason === "max_tokens") {
          text += "\n\n(warning: subagent response was truncated at max_tokens)";
        }
        return text;
      }
      const toolResults: Anthropic.ToolResultBlockParam[] = [];
      let report: string | null = null;
      for (const block of response.content) {
        if (block.type !== "tool_use") {
          continue;
        }
        let outcome: ToolOutcome;
        switch (block.name) {
          case "report_findings":
            report = JSON.stringify(block.input, null, 2);
            outcome = { output: "Findings recorded.", isError: false };
            break;
          case "bash":
            outcome = await handleBashBlock(block);
            break;
          default:
            outcome = { output: `unknown tool: ${block.name}`, isError: true };
        }
        toolResults.push({
          type: "tool_result",
          tool_use_id: block.id,
          content: outcome.output,
          is_error: outcome.isError
        });
      }
      if (report !== null) {
        return report;
      }
      messages.push({ role: "user", content: toolResults });
    }
    return "(subagent hit the turn limit before finishing)";
  }
  ```

  ```csharp C#
  // Ein Subagent: eine kleine verschachtelte Agenten-Schleife mit dem bash-Tool plus report_findings.
  // Subagenten erben die Aufwandsstufe der Hauptschleife.
  JsonSerializerOptions indented = new() { WriteIndented = true };

  async Task<string> RunSubagent(string prompt)
  {
      const string subagentSystem =
          "You are one agent in a larger parallel fan-out, assigned a single subtask. "
          + "Investigate it directly, using bash to check facts rather than guessing, and finish "
          + "by calling report_findings exactly once. Return findings, not narration.";
      List<MessageParam> messages = [new() { Role = Role.User, Content = prompt }];
      for (var turn = 0; turn < maxSubagentTurns; turn++)
      {
          using var deadline = new CancellationTokenSource(TimeSpan.FromSeconds(requestTimeoutSeconds));
          var response = await client.Messages.CreateStreaming(new MessageCreateParams
          {
              Model = model,
              MaxTokens = 64000,
              System = subagentSystem,
              OutputConfig = new OutputConfig { Effort = effort },
              Tools = [bashTool, reportTool],
              Messages = messages,
          }, cancellationToken: deadline.Token).Aggregate();
          messages.Add(new()
          {
              Role = Role.Assistant,
              Content = response.Content.Select(block => new ContentBlockParam(block.Json)).ToList(),
          });
          if (response.StopReason == StopReason.PauseTurn)
          {
              continue;
          }
          if (response.StopReason != StopReason.ToolUse)
          {
              var text = string.Concat(
                  response.Content.Select(block => block.TryPickText(out var textBlock) ? textBlock.Text : ""));
              if (response.StopReason == StopReason.MaxTokens)
              {
                  text += "\n\n(warning: subagent response was truncated at max_tokens)";
              }
              return text;
          }
          List<ContentBlockParam> toolResults = [];
          string? report = null;
          foreach (var block in response.Content)
          {
              if (!block.TryPickToolUse(out var toolUse))
              {
                  continue;
              }
              if (toolUse.Name == "report_findings")
              {
                  report = JsonSerializer.Serialize(toolUse.Input, indented);
              }
              var (output, isError) = toolUse.Name switch
              {
                  "report_findings" => ("Findings recorded.", false),
                  "bash" => await HandleBashBlock(toolUse),
                  _ => ($"unknown tool: {toolUse.Name}", true),
              };
              toolResults.Add(new ToolResultBlockParam(toolUse.ID) { Content = output, IsError = isError });
          }
          if (report is not null)
          {
              return report;
          }
          messages.Add(new() { Role = Role.User, Content = toolResults });
      }
      return "(subagent hit the turn limit before finishing)";
  }
  ```

  ```go Go
  // streamMessage sendet eine Anfrage und sammelt die gestreamten Events zur
  // finalen Nachricht.
  func streamMessage(ctx context.Context, params anthropic.MessageNewParams) (anthropic.Message, error) {
  	ctx, cancel := context.WithTimeout(ctx, requestTimeoutSeconds*time.Second)
  	defer cancel()
  	stream := client.Messages.NewStreaming(ctx, params)
  	defer stream.Close()
  	var message anthropic.Message
  	for stream.Next() {
  		if err := message.Accumulate(stream.Current()); err != nil {
  			return message, err
  		}
  	}
  	return message, stream.Err()
  }

  // textOf verbindet die Textblöcke einer Antwort.
  func textOf(message anthropic.Message) string {
  	var text strings.Builder
  	for _, block := range message.Content {
  		if textBlock, ok := block.AsAny().(anthropic.TextBlock); ok {
  			text.WriteString(textBlock.Text)
  		}
  	}
  	return text.String()
  }

  // runSubagent führt einen Subagenten aus: eine kleine verschachtelte Agenten-Schleife mit dem
  // bash-Tool plus report_findings. Subagenten erben die Aufwandsstufe der Hauptschleife.
  func runSubagent(ctx context.Context, model string, prompt string) (string, error) {
  	const subagentSystem = "You are one agent in a larger parallel fan-out, assigned a single subtask. " +
  		"Investigate it directly, using bash to check facts rather than guessing, and finish " +
  		"by calling report_findings exactly once. Return findings, not narration."
  	messages := []anthropic.MessageParam{anthropic.NewUserMessage(anthropic.NewTextBlock(prompt))}
  	for range maxSubagentTurns {
  		response, err := streamMessage(ctx, anthropic.MessageNewParams{
  			Model:        model,
  			MaxTokens:    64000,
  			System:       []anthropic.TextBlockParam{{Text: subagentSystem}},
  			OutputConfig: anthropic.OutputConfigParam{Effort: effort},
  			Tools:        []anthropic.ToolUnionParam{bashTool, reportTool},
  			Messages:     messages,
  		})
  		if err != nil {
  			return "", err
  		}
  		messages = append(messages, response.ToParam())

  		switch response.StopReason {
  		case anthropic.StopReasonPauseTurn:
  			continue
  		case anthropic.StopReasonMaxTokens:
  			return textOf(response) + "\n\n(warning: subagent response was truncated at max_tokens)", nil
  		case anthropic.StopReasonToolUse:
  		default:
  			return textOf(response), nil
  		}

  		var toolResults []anthropic.ContentBlockParamUnion
  		var report string
  		var reportRecorded bool
  		for _, block := range response.Content {
  			toolUse, ok := block.AsAny().(anthropic.ToolUseBlock)
  			if !ok {
  				continue
  			}
  			var output string
  			var isError bool
  			switch toolUse.Name {
  			case "report_findings":
  				report = string(toolUse.Input)
  				var pretty bytes.Buffer
  				if err := json.Indent(&pretty, toolUse.Input, "", "  "); err == nil {
  					report = pretty.String()
  				}
  				reportRecorded = true
  				output = "Findings recorded."
  			case "bash":
  				output, isError = handleBashBlock(ctx, toolUse)
  			default:
  				output, isError = "unknown tool: "+toolUse.Name, true
  			}
  			toolResults = append(toolResults, anthropic.NewToolResultBlock(toolUse.ID, output, isError))
  		}
  		if reportRecorded {
  			return report, nil
  		}
  		messages = append(messages, anthropic.NewUserMessage(toolResults...))
  	}
  	return "(subagent hit the turn limit before finishing)", nil
  }
  ```

  ```java Java
  static final String SUBAGENT_SYSTEM = """
          You are one agent in a larger parallel fan-out, assigned a single subtask. \
          Investigate it directly, using bash to check facts rather than guessing, and finish \
          by calling report_findings exactly once. Return findings, not narration.""";

  // Ein Subagent: eine kleine verschachtelte Agenten-Schleife mit dem bash-Tool plus report_findings.
  // Subagenten erben die Aufwandsstufe der Hauptschleife.
  String runSubagent(Model model, String prompt) throws InterruptedException {
      List<MessageParam> messages = new ArrayList<>();
      messages.add(MessageParam.builder().role(MessageParam.Role.USER).content(prompt).build());
      for (int turn = 0; turn < MAX_SUBAGENT_TURNS; turn++) {
          MessageCreateParams params = MessageCreateParams.builder()
                  .model(model)
                  .maxTokens(64000L)
                  .system(SUBAGENT_SYSTEM)
                  .outputConfig(OutputConfig.builder().effort(EFFORT).build())
                  .addTool(BASH_TOOL)
                  .addTool(REPORT_TOOL)
                  .messages(messages)
                  .build();
          MessageAccumulator accumulator = MessageAccumulator.create();
          try (var stream = client.messages().createStreaming(params, REQUEST_OPTIONS)) {
              stream.stream().forEach(accumulator::accumulate);
          }
          Message response = accumulator.message();
          messages.add(response.toParam());
          StopReason stopReason = response.stopReason().orElse(null);
          if (StopReason.PAUSE_TURN.equals(stopReason)) {
              continue;
          }
          if (!StopReason.TOOL_USE.equals(stopReason)) {
              String text = response.content().stream()
                      .flatMap(block -> block.text().stream())
                      .map(TextBlock::text)
                      .collect(Collectors.joining());
              if (StopReason.MAX_TOKENS.equals(stopReason)) {
                  text += "\n\n(warning: subagent response was truncated at max_tokens)";
              }
              return text;
          }
          List<ToolUseBlock> toolUses = response.content().stream()
                  .flatMap(block -> block.toolUse().stream())
                  .toList();
          List<ContentBlockParam> toolResults = new ArrayList<>();
          String report = null;
          for (ToolUseBlock toolUse : toolUses) {
              ToolOutput result = switch (toolUse.name()) {
                  case "report_findings" -> {
                      report = toolUse._input().convert(JsonNode.class).toPrettyString();
                      yield new ToolOutput("Findings recorded.", false);
                  }
                  case "bash" -> handleBashBlock(toolUse);
                  default -> new ToolOutput("unknown tool: " + toolUse.name(), true);
              };
              toolResults.add(ContentBlockParam.ofToolResult(ToolResultBlockParam.builder()
                      .toolUseId(toolUse.id())
                      .content(result.output())
                      .isError(result.isError())
                      .build()));
          }
          if (report != null) {
              return report;
          }
          messages.add(MessageParam.builder()
                  .role(MessageParam.Role.USER)
                  .contentOfBlockParams(toolResults)
                  .build());
      }
      return "(subagent hit the turn limit before finishing)";
  }
  ```

  ```php PHP
  /**
   * Consume a message stream and assemble the final assistant turn from its events:
   * the full content-block list plus the stop reason, equivalent to what a
   * non-streaming create call returns.
   */
  function drainMessageStream(iterable $events): array
  {
      $stringValue = fn ($value) => $value instanceof \BackedEnum ? $value->value : $value;
      $blocks = [];
      $jsonBuffers = [];
      $stopReason = null;
      foreach ($events as $event) {
          $type = $stringValue($event->type);
          if ($type === 'content_block_start') {
              $blocks[$event->index] = $event->contentBlock;
              $jsonBuffers[$event->index] = '';
          } elseif ($type === 'content_block_delta') {
              $block = $blocks[$event->index];
              $delta = $event->delta;
              $deltaType = $stringValue($delta->type);
              if ($deltaType === 'text_delta') {
                  $blocks[$event->index] = $block->withText($block->text . $delta->text);
              } elseif ($deltaType === 'input_json_delta') {
                  $jsonBuffers[$event->index] .= $delta->partialJSON;
              } elseif ($deltaType === 'thinking_delta') {
                  $blocks[$event->index] = $block->withThinking($block->thinking . $delta->thinking);
              } elseif ($deltaType === 'signature_delta') {
                  $blocks[$event->index] = $block->withSignature($delta->signature);
              }
          } elseif ($type === 'message_delta') {
              $stopReason = $stringValue($event->delta->stopReason);
          }
      }
      foreach ($jsonBuffers as $index => $buffer) {
          if ($buffer !== '' && $blocks[$index] instanceof \Anthropic\Messages\ToolUseBlock) {
              $decoded = json_decode($buffer, true);
              $blocks[$index] = $blocks[$index]->withInput(is_array($decoded) ? $decoded : []);
          }
      }
      return [array_values($blocks), $stopReason];
  }

  /**
   * One subagent: a small nested agent loop with the bash tool plus report_findings.
   * Subagents inherit the main loop's effort level.
   */
  function runSubagent(Client $client, string $model, string $prompt): string
  {
      $subagentSystem =
          'You are one agent in a larger parallel fan-out, assigned a single subtask. '
          . 'Investigate it directly, using bash to check facts rather than guessing, and finish '
          . 'by calling report_findings exactly once. Return findings, not narration.';
      $messages = [['role' => 'user', 'content' => $prompt]];
      for ($turn = 0; $turn < MAX_SUBAGENT_TURNS; $turn++) {
          $stream = $client->messages->createStream(
              model: $model,
              maxTokens: 64000,
              system: $subagentSystem,
              outputConfig: ['effort' => EFFORT],
              tools: [BASH_TOOL, REPORT_TOOL],
              messages: $messages,
              requestOptions: ['timeout' => REQUEST_TIMEOUT_SECONDS],
          );
          [$content, $stopReason] = drainMessageStream($stream);
          $messages[] = ['role' => 'assistant', 'content' => $content];
          if ($stopReason === 'pause_turn') {
              continue;
          }
          if ($stopReason !== 'tool_use') {
              $text = '';
              foreach ($content as $block) {
                  if ($block instanceof \Anthropic\Messages\TextBlock) {
                      $text .= $block->text;
                  }
              }
              if ($stopReason === 'max_tokens') {
                  $text .= "\n\n(warning: subagent response was truncated at max_tokens)";
              }
              return $text;
          }
          $report = null;
          $toolResults = [];
          foreach ($content as $block) {
              if (!$block instanceof \Anthropic\Messages\ToolUseBlock) {
                  continue;
              }
              if ($block->name === 'report_findings') {
                  $report = json_encode($block->input, JSON_PRETTY_PRINT);
                  $output = 'Findings recorded.';
                  $isError = false;
              } elseif ($block->name === 'bash') {
                  [$output, $isError] = handleBashBlock($block);
              } else {
                  $output = "unknown tool: {$block->name}";
                  $isError = true;
              }
              $toolResults[] = [
                  'type' => 'tool_result',
                  'tool_use_id' => $block->id,
                  'content' => $output,
                  'is_error' => $isError,
              ];
          }
          if ($report !== null) {
              return $report;
          }
          $messages[] = ['role' => 'user', 'content' => $toolResults];
      }
      return '(subagent hit the turn limit before finishing)';
  }
  ```

  ```ruby Ruby
  # Ein Subagent: eine kleine verschachtelte Agenten-Schleife mit dem bash-Tool plus report_findings.
  # Subagenten erben das Effort-Level der Hauptschleife.
  def run_subagent(model, prompt)
    subagent_system =
      "You are one agent in a larger parallel fan-out, assigned a single subtask. " \
      "Investigate it directly, using bash to check facts rather than guessing, and finish " \
      "by calling report_findings exactly once. Return findings, not narration."
    messages = [{role: "user", content: prompt}]
    MAX_SUBAGENT_TURNS.times do
      stream = CLIENT.messages.stream(
        model: model,
        max_tokens: 64_000,
        system_: subagent_system,
        output_config: {effort: EFFORT},
        tools: [BASH_TOOL, REPORT_TOOL],
        messages: messages,
        request_options: {timeout: REQUEST_TIMEOUT_SECONDS}
      )
      response = stream.accumulated_message
      messages << {role: "assistant", content: assistant_content_param(response.content)}
      next if response.stop_reason == :pause_turn

      unless response.stop_reason == :tool_use
        text = response.content.select { |block| block.type == :text }.map(&:text).join
        text += "\n\n(warning: subagent response was truncated at max_tokens)" if response.stop_reason == :max_tokens
        return text
      end

      report = nil
      tool_results = []
      response.content.each do |block|
        next unless block.type == :tool_use

        input = parse_tool_input(block.input)
        case block.name
        when "report_findings"
          report = JSON.pretty_generate(input)
          output, is_error = "Findings recorded.", false
        when "bash"
          output, is_error = handle_bash_block(block)
        else
          output, is_error = "unknown tool: #{block.name}", true
        end
        tool_results << {
          type: "tool_result",
          tool_use_id: block.id,
          content: output,
          is_error: is_error
        }
      end
      return report unless report.nil?

      messages << {role: "user", content: tool_results}
    end
    "(subagent hit the turn limit before finishing)"
  end
  ```
</CodeGroup>

## Ergebnisse protokollieren, damit erneute Läufe fortsetzen

Ein Fan-out, der Dutzende Subagenten erzeugt, ist teuer, wenn man ihn von Grund auf neu startet. Ein kleines inhaltsadressiertes Journal macht ihn idempotent: Bevor ein Subagent losgeschickt wird, wird der SHA-256-Hash seines Prompts in einer lokalen JSON-Datei nachgeschlagen, und falls ein aufgezeichnetes Ergebnis existiert, wird dieses zurückgegeben. Unterbrich den Lauf, starte ihn erneut, und nur die Teilaufgaben, die nie fertig wurden, werden neu berechnet. Das Journal dedupliziert über Läufe hinweg, nicht innerhalb einer einzelnen Fan-out-Welle; lösche die Journal-Datei, um neu zu beginnen.

<CodeGroup exclude="shell:cURL, shell:CLI">
  ```python Python
  _journal_lock = threading.Lock()


  def _load_journal() -> dict:
      try:
          with open(JOURNAL_PATH) as file:
              return json.load(file) or {}
      except (OSError, json.JSONDecodeError):
          return {}


  def journaled(prompt: str, compute) -> str:
      """Return a cached result for this exact prompt, or compute and persist it. This
      makes the fan-out resumable: interrupt the run, rerun it, and only the subtasks
      that never finished are recomputed. Delete the journal file to start fresh."""
      key = hashlib.sha256(prompt.encode()).hexdigest()
      cached = _load_journal().get(key)
      if cached is not None:
          print(f"[journal] cache hit for {key[:12]}", file=sys.stderr)
          return cached
      result = compute()
      try:
          with _journal_lock:  # fan-out writes from many threads
              journal = _load_journal()
              journal[key] = result
              temp = f"{JOURNAL_PATH}.tmp"
              with open(temp, "w") as file:
                  json.dump(journal, file)
              os.replace(temp, JOURNAL_PATH)  # atomic on POSIX and Windows
      # Ein fehlgeschlagener Journal-Schreibvorgang darf das berechnete Ergebnis nicht verlieren
      except OSError as error:
          print(f"[journal] write failed: {error}", file=sys.stderr)
      return result
  ```

  ```typescript TypeScript
  let journalWriteChain = Promise.resolve();

  async function loadJournal(): Promise<Record<string, string>> {
    try {
      return JSON.parse(await readFile(JOURNAL_PATH, "utf8")) ?? {};
    } catch (error) {
      if ((error as NodeJS.ErrnoException).code !== "ENOENT") {
        console.error(`[journal] discarding unreadable journal: ${error}`);
      }
      return {};
    }
  }

  // Gibt ein gecachtes Ergebnis für genau diesen Prompt zurück oder berechnet und speichert es. Das
  // macht das Fan-out fortsetzbar: Lauf unterbrechen, erneut starten, und nur die Teilaufgaben,
  // die nie fertig wurden, werden neu berechnet. Lösche die Journal-Datei für einen Neustart.
  async function journaled(prompt: string, compute: () => Promise<string>): Promise<string> {
    const key = createHash("sha256").update(prompt).digest("hex");
    const cached = (await loadJournal())[key];
    if (cached !== undefined) {
      console.error(`[journal] cache hit for ${key.slice(0, 12)}`);
      return cached;
    }
    const result = await compute();
    // Verkettet Schreibvorgänge, damit parallele Subagenten sich nicht gegenseitig die Einträge überschreiben.
    // Die Kette wird settled gehalten, damit ein fehlgeschlagener Schreibvorgang spätere nicht vergiftet.
    await (journalWriteChain = journalWriteChain
      .then(async () => {
        const journal = await loadJournal();
        journal[key] = result;
        const temp = `${JOURNAL_PATH}.tmp`;
        await writeFile(temp, JSON.stringify(journal));
        await rename(temp, JOURNAL_PATH);
      })
      .catch((error) => console.error(`[journal] write failed: ${error}`)));
    return result;
  }
  ```

  ```csharp C#
  SemaphoreSlim journalLock = new(1, 1);

  async Task<Dictionary<string, string>> LoadJournal()
  {
      try
      {
          return JsonSerializer.Deserialize<Dictionary<string, string>>(await File.ReadAllTextAsync(journalPath)) ?? [];
      }
      catch (Exception error) when (error is IOException or UnauthorizedAccessException or JsonException)
      {
          return [];
      }
  }

  // Gib ein gecachtes Ergebnis für genau diesen Prompt zurück oder berechne und speichere es. Das
  // macht das Fan-out fortsetzbar: unterbrich den Lauf, starte ihn neu, und nur die Subtasks,
  // die nie fertig wurden, werden neu berechnet. Lösche die Journal-Datei für einen Neustart.
  async Task<string> Journaled(string prompt, Func<Task<string>> compute)
  {
      var key = Convert.ToHexStringLower(SHA256.HashData(Encoding.UTF8.GetBytes(prompt)));
      if ((await LoadJournal()).TryGetValue(key, out var cached))
      {
          Console.Error.WriteLine($"[journal] cache hit for {key[..12]}");
          return cached;
      }
      var result = await compute();
      await journalLock.WaitAsync(); // fan-out writes from many tasks
      try
      {
          var journal = await LoadJournal();
          journal[key] = result;
          var temp = journalPath + ".tmp";
          await File.WriteAllTextAsync(temp, JsonSerializer.Serialize(journal));
          File.Move(temp, journalPath, overwrite: true);
      }
      catch (Exception error) when (error is IOException or UnauthorizedAccessException or NotSupportedException)
      {
          // Das Journal ist Best-Effort; verwirf niemals ein berechnetes Ergebnis.
          Console.Error.WriteLine($"[journal] write failed: {error.Message}");
      }
      finally
      {
          journalLock.Release();
      }
      return result;
  }
  ```

  ```go Go
  var journalMutex sync.Mutex

  func loadJournal() map[string]string {
  	data, err := os.ReadFile(journalPath)
  	if err != nil {
  		return map[string]string{}
  	}
  	var journal map[string]string
  	if err := json.Unmarshal(data, &journal); err != nil || journal == nil {
  		return map[string]string{}
  	}
  	return journal
  }

  // journaled gibt ein gecachtes Ergebnis für genau diesen Prompt zurück oder berechnet und
  // persistiert es. Das macht das Fan-out fortsetzbar: unterbrich den Lauf, starte ihn neu, und nur
  // nie abgeschlossene Subtasks werden neu berechnet. Lösche die Journal-Datei für einen Neustart.
  func journaled(prompt string, compute func() (string, error)) (string, error) {
  	sum := sha256.Sum256([]byte(prompt))
  	key := hex.EncodeToString(sum[:])
  	if cached, ok := loadJournal()[key]; ok {
  		fmt.Fprintf(os.Stderr, "[journal] cache hit for %s\n", key[:12])
  		return cached, nil
  	}
  	result, err := compute()
  	if err != nil {
  		return "", err
  	}
  	journalMutex.Lock() // fan-out writes from many goroutines
  	defer journalMutex.Unlock()
  	journal := loadJournal()
  	journal[key] = result
  	data, _ := json.Marshal(journal)
  	temp := journalPath + ".tmp"
  	if err := os.WriteFile(temp, data, 0o644); err != nil {
  		fmt.Fprintf(os.Stderr, "[journal] write failed: %s\n", err)
  	} else if err := os.Rename(temp, journalPath); err != nil {
  		fmt.Fprintf(os.Stderr, "[journal] write failed: %s\n", err)
  		_ = os.Remove(temp)
  	}
  	return result, nil
  }
  ```

  ```java Java
  static final ObjectMapper JOURNAL_MAPPER = new ObjectMapper();
  static final ReentrantLock JOURNAL_LOCK = new ReentrantLock();

  Map<String, String> loadJournal() {
      try {
          return Objects.requireNonNullElseGet(
                  JOURNAL_MAPPER.readValue(Files.readString(JOURNAL_PATH), new TypeReference<HashMap<String, String>>() {}),
                  HashMap::new);
      } catch (IOException _) {
          return new HashMap<>();
      }
  }

  // Gib ein gecachtes Ergebnis für genau diesen Prompt zurück oder berechne und speichere es. Das
  // macht das Fan-out fortsetzbar: unterbrich den Lauf, starte ihn neu, und nur die Teilaufgaben,
  // die nie abgeschlossen wurden, werden neu berechnet. Lösche die Journal-Datei für einen Neustart.
  String journaled(String prompt, Callable<String> compute) throws Exception {
      var digest = MessageDigest.getInstance("SHA-256").digest(prompt.getBytes(StandardCharsets.UTF_8));
      String key = HexFormat.of().formatHex(digest);
      String cached = loadJournal().get(key);
      if (cached != null) {
          System.err.println("[journal] cache hit for " + key.substring(0, 12));
          return cached;
      }
      String result = compute.call();
      JOURNAL_LOCK.lock(); // fan-out writes from many threads
      try {
          Map<String, String> journal = loadJournal();
          journal.put(key, result);
          Path temp = JOURNAL_PATH.resolveSibling(JOURNAL_PATH.getFileName() + ".tmp");
          Files.writeString(temp, JOURNAL_MAPPER.writeValueAsString(journal));
          Files.move(temp, JOURNAL_PATH, StandardCopyOption.REPLACE_EXISTING, StandardCopyOption.ATOMIC_MOVE);
      } catch (IOException error) {
          // Das Journal ist Best-Effort; verwirf niemals ein berechnetes Ergebnis.
          System.err.println("[journal] write failed: " + error);
      } finally {
          JOURNAL_LOCK.unlock();
      }
      return result;
  }
  ```

  ```php PHP
  function loadJournal(): array
  {
      $raw = @file_get_contents(JOURNAL_PATH);
      if ($raw === false) {
          return [];
      }
      $decoded = json_decode($raw, true);
      return is_array($decoded) ? $decoded : [];
  }

  /**
   * Return a cached result for this exact prompt, or compute and persist it. This
   * makes the fan-out resumable: interrupt the run, rerun it, and only the subtasks
   * that never finished are recomputed. Delete the journal file to start fresh.
   */
  function journaled(string $prompt, callable $compute): string
  {
      $key = hash('sha256', $prompt);
      $journal = loadJournal();
      if (array_key_exists($key, $journal)) {
          fwrite(STDERR, '[journal] cache hit for ' . substr($key, 0, 12) . "\n");
          return $journal[$key];
      }
      $result = $compute();
      $journal = loadJournal();
      $journal[$key] = $result;
      $temp = JOURNAL_PATH . '.tmp';
      $encoded = json_encode($journal, JSON_INVALID_UTF8_SUBSTITUTE);
      if ($encoded === false || @file_put_contents($temp, $encoded) === false || !@rename($temp, JOURNAL_PATH)) {
          fwrite(STDERR, '[journal] write failed: ' . (error_get_last()['message'] ?? json_last_error_msg()) . "\n");
          @unlink($temp);
      }
      return $result;
  }
  ```

  ```ruby Ruby
  JOURNAL_LOCK = Mutex.new

  def load_journal
    JSON.parse(File.read(JOURNAL_PATH)) || {}
  rescue SystemCallError, JSON::ParserError
    {}
  end

  # Gib ein gecachtes Ergebnis für genau diesen Prompt zurück oder berechne und speichere es. Das
  # macht das Fan-out fortsetzbar: unterbrich den Lauf, starte ihn erneut, und nur die Teilaufgaben,
  # die nie fertig wurden, werden neu berechnet. Lösche die Journal-Datei für einen Neustart.
  def journaled(prompt)
    key = Digest::SHA256.hexdigest(prompt)
    cached = load_journal[key]
    unless cached.nil?
      warn "[journal] cache hit for #{key[0, 12]}"
      return cached
    end
    result = yield
    begin
      JOURNAL_LOCK.synchronize do # fan-out writes from many threads
        journal = load_journal
        journal[key] = result
        temp = "#{JOURNAL_PATH}.tmp"
        File.write(temp, JSON.generate(journal))
        File.rename(temp, JOURNAL_PATH)
      end
    rescue SystemCallError => error # the journal is best-effort; never discard a computed result
      warn "[journal] write failed: #{error}"
    end
    result
  end
  ```
</CodeGroup>

## Fan-out, dann verifizieren

Der Fan-out akzeptiert bis zu `MAX_TOTAL_SUBTASKS` Prompts, führt sie über das Journal mit höchstens `MAX_CONCURRENT` gleichzeitig laufenden aus (sequenziell in der PHP-Portierung) und isoliert Fehler, sodass ein defekter Subagent zu einem Fehler-String degradiert, statt den Lauf zu beenden. Sobald die erste Welle abgeschlossen ist, nutzt eine zweite Welle denselben Subagenten-Pfad, um zu versuchen, jedes Ergebnis zu widerlegen: Jeder Verifizierer leitet die Behauptungen erneut aus der Quelle ab und entscheidet im Zweifel standardmäßig auf widerlegt. Sowohl das ursprüngliche Ergebnis als auch sein Urteil werden an den Orchestrator zurückgegeben, damit er beide gemeinsam abwägen kann.

<CodeGroup exclude="shell:cURL, shell:CLI">
  ```python Python
  def normalize_subtasks(raw) -> list[str]:
      """Accept the subtasks input in whatever shape the model emits: an array, the array
      JSON-encoded as a single string, or a newline-separated list."""
      if isinstance(raw, str):
          try:
              raw = json.loads(raw)
          except json.JSONDecodeError:
              raw = raw.splitlines() if "\n" in raw else [raw]
      if not isinstance(raw, list):
          return []
      return [task.strip() for task in raw if isinstance(task, str) and task.strip()]


  def verify_prompt_for(subtask: str, result: str) -> str:
      return (
          "Adversarially verify the subagent result below: try to REFUTE it. Re-derive the "
          "claims yourself with bash rather than trusting the result, and look for evidence "
          "that contradicts them. Default to refuted if uncertain. Call report_findings with "
          "summary 'refuted: <why>' or 'confirmed: <why>', citing the file:line or command "
          "output that decided it.\n\n"
          f"""Subtask: {subtask}

  Result to verify:
  {result}"""
      )


  def run_workflow(model: str, raw_subtasks) -> tuple[str, bool]:
      """Run subtasks as parallel subagents, then run a second verification wave over
      the results, and return both. MAX_TOTAL_SUBTASKS bounds how many the model can
      queue; MAX_CONCURRENT bounds how many run at once."""
      all_subtasks = normalize_subtasks(raw_subtasks)
      subtasks = all_subtasks[:MAX_TOTAL_SUBTASKS]
      dropped = len(all_subtasks) - len(subtasks)
      if not subtasks:
          return "Workflow error: no usable subtasks were provided.", True
      print(f"[workflow] fanning out {len(subtasks)} agents", file=sys.stderr)

      def run_one(prompt: str) -> str:
          try:
              return journaled(prompt, lambda: run_subagent(model, prompt))
          # Ein fehlgeschlagener Subagent darf den Lauf nicht beenden
          except Exception as error:
              return f"(subagent failed: {type(error).__name__}: {error})"

      with concurrent.futures.ThreadPoolExecutor(max_workers=MAX_CONCURRENT) as pool:
          results = list(pool.map(run_one, subtasks))
          print(f"[workflow] verifying {len(results)} results", file=sys.stderr)
          verify_prompts = [
              verify_prompt_for(task, result) for task, result in zip(subtasks, results)
          ]
          verdicts = list(pool.map(run_one, verify_prompts))

      joined = "\n\n".join(
          f"""[agent {index + 1}: {task}]
  {result}

  [verify {index + 1}]
  {verdict}"""
          for index, (task, result, verdict) in enumerate(
              zip(subtasks, results, verdicts)
          )
      )
      if dropped > 0:
          joined = (
              f"(note: {dropped} subtasks beyond MAX_TOTAL_SUBTASKS={MAX_TOTAL_SUBTASKS} were not "
              "run; rerun them in a follow-up Workflow call)\n\n" + joined
          )
      return joined, False
  ```

  ```typescript TypeScript
  // Akzeptiert die subtasks-Eingabe in jeder Form, die das Modell ausgibt: ein Array, das Array
  // JSON-kodiert als einzelner String oder eine durch Zeilenumbrüche getrennte Liste.
  function normalizeSubtasks(raw: unknown): string[] {
    let value = raw;
    if (typeof raw === "string") {
      try {
        value = JSON.parse(raw);
      } catch {
        value = raw.includes("\n") ? raw.split("\n") : [raw];
      }
    }
    if (!Array.isArray(value)) {
      return [];
    }
    return value
      .filter((task) => typeof task === "string")
      .map((task) => task.trim())
      .filter((task) => task.length > 0);
  }

  function verifyPromptFor(subtask: string, result: string): string {
    const instructions =
      "Adversarially verify the subagent result below: try to REFUTE it. Re-derive the " +
      "claims yourself with bash rather than trusting the result, and look for evidence " +
      "that contradicts them. Default to refuted if uncertain. Call report_findings with " +
      "summary 'refuted: <why>' or 'confirmed: <why>', citing the file:line or command " +
      "output that decided it.";
    return `${instructions}

  Subtask: ${subtask}

  Result to verify:
  ${result}`;
  }

  // Map mit Nebenläufigkeitslimit: höchstens `limit` Aufgaben laufen gleichzeitig.
  async function mapWithLimit<In, Out>(
    items: readonly In[],
    limit: number,
    task: (item: In) => Promise<Out>
  ): Promise<Out[]> {
    const results = new Array<Out>(items.length);
    let cursor = 0;
    const workers = Array.from({ length: Math.min(limit, items.length) }, async () => {
      while (cursor < items.length) {
        const index = cursor++;
        results[index] = await task(items[index]);
      }
    });
    await Promise.all(workers);
    return results;
  }

  // Führt Teilaufgaben als parallele Subagenten aus, lässt dann eine zweite Verifikationswelle über
  // die Ergebnisse laufen und gibt beides zurück. MAX_TOTAL_SUBTASKS begrenzt, wie viele das Modell
  // einreihen kann; MAX_CONCURRENT begrenzt, wie viele gleichzeitig laufen.
  async function runWorkflow(model: string, rawSubtasks: unknown): Promise<ToolOutcome> {
    const allSubtasks = normalizeSubtasks(rawSubtasks);
    const subtasks = allSubtasks.slice(0, MAX_TOTAL_SUBTASKS);
    const dropped = allSubtasks.length - subtasks.length;
    if (subtasks.length === 0) {
      return { output: "Workflow error: no usable subtasks were provided.", isError: true };
    }
    console.error(`[workflow] fanning out ${subtasks.length} agents`);

    const runOne = async (prompt: string): Promise<string> => {
      try {
        return await journaled(prompt, () => runSubagent(model, prompt));
      } catch (error) {
        // Isolationsgrenze: ein fehlerhafter Subagent soll nicht den ganzen Lauf beenden.
        const reason =
          error instanceof Error ? `${error.name}: ${error.message}` : String(error);
        return `(subagent failed: ${reason})`;
      }
    };

    const results = await mapWithLimit(subtasks, MAX_CONCURRENT, runOne);
    console.error(`[workflow] verifying ${results.length} results`);
    const verifyPrompts = subtasks.map((task, index) => verifyPromptFor(task, results[index]));
    const verdicts = await mapWithLimit(verifyPrompts, MAX_CONCURRENT, runOne);

    const report = subtasks
      .map(
        (task, index) => `[agent ${index + 1}: ${task}]
  ${results[index]}

  [verify ${index + 1}]
  ${verdicts[index]}`
      )
      .join("\n\n");
    const droppedNote =
      dropped > 0
        ? `(note: ${dropped} subtasks beyond MAX_TOTAL_SUBTASKS=${MAX_TOTAL_SUBTASKS} were ` +
          "not run; rerun them in a follow-up Workflow call)\n\n"
        : "";
    return { output: droppedNote + report, isError: false };
  }
  ```

  ```csharp C#
  // Akzeptiere die subtasks-Eingabe in jeder Form, die das Modell ausgibt: ein Array, das Array
  // JSON-kodiert als einzelner String oder eine durch Zeilenumbrüche getrennte Liste.
  List<string> NormalizeSubtasks(JsonElement raw)
  {
      IEnumerable<string?> tasks = raw.ValueKind switch
      {
          JsonValueKind.Array => raw.EnumerateArray()
              .Where(item => item.ValueKind == JsonValueKind.String)
              .Select(item => item.GetString()),
          JsonValueKind.String => ParseSubtaskString(raw.GetString()!),
          _ => [],
      };
      return [.. tasks.OfType<string>().Select(task => task.Trim()).Where(task => task.Length > 0)];
  }

  IEnumerable<string?> ParseSubtaskString(string single)
  {
      try
      {
          return JsonSerializer.Deserialize<List<string?>>(single) ?? [];
      }
      catch (JsonException)
      {
          return single.Split('\n');
      }
  }

  string VerifyPromptFor(string subtask, string result) => $"""
      Adversarially verify the subagent result below: try to REFUTE it. Re-derive the claims yourself with bash rather than trusting the result, and look for evidence that contradicts them. Default to refuted if uncertain. Call report_findings with summary 'refuted: <why>' or 'confirmed: <why>', citing the file:line or command output that decided it.

      Subtask: {subtask}

      Result to verify:
      {result}
      """;

  // Führe Subtasks als parallele Subagenten aus, dann eine zweite Verifizierungswelle über
  // die Ergebnisse, und gib beides zurück. maxTotalSubtasks begrenzt, wie viele das Modell
  // einreihen kann; maxConcurrent begrenzt, wie viele gleichzeitig laufen.
  async Task<(string Output, bool IsError)> RunWorkflow(JsonElement rawSubtasks)
  {
      var allSubtasks = NormalizeSubtasks(rawSubtasks);
      List<string> subtasks = [.. allSubtasks.Take(maxTotalSubtasks)];
      var dropped = allSubtasks.Count - subtasks.Count;
      if (subtasks.Count == 0)
      {
          return ("Workflow error: no usable subtasks were provided.", true);
      }
      Console.Error.WriteLine($"[workflow] fanning out {subtasks.Count} agents");

      using SemaphoreSlim gate = new(maxConcurrent);
      async Task<string> RunOne(string prompt)
      {
          await gate.WaitAsync();
          try
          {
              return await Journaled(prompt, () => RunSubagent(prompt));
          }
          catch (Exception error)
          {
              // Isolationsgrenze: ein fehlerhafter Subagent sollte den Lauf nicht beenden.
              return $"(subagent failed: {error.GetType().Name}: {error.Message})";
          }
          finally
          {
              gate.Release();
          }
      }

      var results = await Task.WhenAll(subtasks.Select(RunOne));
      Console.Error.WriteLine($"[workflow] verifying {results.Length} results");
      var verdicts = await Task.WhenAll(subtasks.Zip(results, VerifyPromptFor).Select(RunOne));

      var joined = string.Join("\n\n", subtasks.Select((task, index) => $"""
          [agent {index + 1}: {task}]
          {results[index]}

          [verify {index + 1}]
          {verdicts[index]}
          """));
      if (dropped > 0)
      {
          joined = $"""
              (note: {dropped} subtasks beyond maxTotalSubtasks={maxTotalSubtasks} were not run; rerun them in a follow-up Workflow call)

              {joined}
              """;
      }
      return (joined, false);
  }
  ```

  ```go Go
  // normalizeSubtasks akzeptiert die Subtasks-Eingabe in jeder Form, die das Modell ausgibt: ein
  // Array, das Array als einzelner JSON-kodierter String oder eine zeilengetrennte Liste.
  func normalizeSubtasks(raw json.RawMessage) []string {
  	var tasks []string
  	if err := json.Unmarshal(raw, &tasks); err != nil {
  		var single string
  		if err := json.Unmarshal(raw, &single); err != nil {
  			return nil
  		}
  		if err := json.Unmarshal([]byte(single), &tasks); err != nil {
  			tasks = strings.Split(single, "\n")
  		}
  	}
  	cleaned := make([]string, 0, len(tasks))
  	for _, task := range tasks {
  		if trimmed := strings.TrimSpace(task); trimmed != "" {
  			cleaned = append(cleaned, trimmed)
  		}
  	}
  	return cleaned
  }

  func verifyPromptFor(subtask, result string) string {
  	return "Adversarially verify the subagent result below: try to REFUTE it. Re-derive the " +
  		"claims yourself with bash rather than trusting the result, and look for evidence " +
  		"that contradicts them. Default to refuted if uncertain. Call report_findings with " +
  		"summary 'refuted: <why>' or 'confirmed: <why>', citing the file:line or command " +
  		"output that decided it.\n\n" +
  		fmt.Sprintf(`Subtask: %s

  Result to verify:
  %s`, subtask, result)
  }

  // mapWithLimit führt task über items aus, mit höchstens limit Goroutinen gleichzeitig.
  func mapWithLimit(items []string, limit int, task func(string) string) []string {
  	results := make([]string, len(items))
  	semaphore := make(chan struct{}, limit)
  	var waitGroup sync.WaitGroup
  	for index, item := range items {
  		semaphore <- struct{}{}
  		waitGroup.Go(func() {
  			defer func() { <-semaphore }()
  			results[index] = task(item)
  		})
  	}
  	waitGroup.Wait()
  	return results
  }

  // runWorkflow führt Subtasks als parallele Subagenten aus, dann eine zweite Verifizierungswelle
  // über die Ergebnisse, und gibt beides zurück. maxTotalSubtasks begrenzt, wie viele das Modell
  // einreihen kann; maxConcurrent begrenzt, wie viele gleichzeitig laufen.
  func runWorkflow(ctx context.Context, model string, rawSubtasks json.RawMessage) (string, bool) {
  	allSubtasks := normalizeSubtasks(rawSubtasks)
  	subtasks := allSubtasks[:min(len(allSubtasks), maxTotalSubtasks)]
  	dropped := len(allSubtasks) - len(subtasks)
  	if len(subtasks) == 0 {
  		return "Workflow error: no usable subtasks were provided.", true
  	}
  	fmt.Fprintf(os.Stderr, "[workflow] fanning out %d agents\n", len(subtasks))

  	runOne := func(prompt string) string {
  		report, err := journaled(prompt, func() (string, error) { return runSubagent(ctx, model, prompt) })
  		if err != nil {
  			// Isolationsgrenze: ein fehlerhafter Subagent soll nicht den ganzen Lauf beenden.
  			return fmt.Sprintf("(subagent failed: %s)", err)
  		}
  		return report
  	}

  	results := mapWithLimit(subtasks, maxConcurrent, runOne)
  	fmt.Fprintf(os.Stderr, "[workflow] verifying %d results\n", len(results))
  	verifyPrompts := make([]string, len(subtasks))
  	for index, task := range subtasks {
  		verifyPrompts[index] = verifyPromptFor(task, results[index])
  	}
  	verdicts := mapWithLimit(verifyPrompts, maxConcurrent, runOne)

  	sections := make([]string, len(subtasks))
  	for index, task := range subtasks {
  		sections[index] = fmt.Sprintf(`[agent %d: %s]
  %s

  [verify %d]
  %s`,
  			index+1, task, results[index], index+1, verdicts[index])
  	}
  	joined := strings.Join(sections, "\n\n")
  	if dropped > 0 {
  		joined = fmt.Sprintf("(note: %d subtasks beyond maxTotalSubtasks=%d were not run; "+
  			"rerun them in a follow-up Workflow call)\n\n", dropped, maxTotalSubtasks) + joined
  	}
  	return joined, false
  }
  ```

  ```java Java
  // Akzeptiere die subtasks-Eingabe in jeder Form, die das Modell ausgibt: ein Array, das Array
  // als einzelner String JSON-kodiert oder eine durch Zeilenumbrüche getrennte Liste.
  List<String> normalizeSubtasks(JsonValue rawSubtasks) {
      List<String> tasks = switch (rawSubtasks) {
          case JsonArray array -> array.values().stream()
                  .map(item -> item instanceof JsonString text ? text.value() : item.toString())
                  .toList();
          case JsonString string -> parseSubtaskString(string.value());
          default -> List.of();
      };
      return tasks.stream()
              .filter(Objects::nonNull)
              .map(String::strip)
              .filter(task -> !task.isEmpty())
              .toList();
  }

  List<String> parseSubtaskString(String encoded) {
      try {
          String[] parsed = new ObjectMapper().readValue(encoded, String[].class);
          return parsed == null ? List.of() : Arrays.asList(parsed);
      } catch (JsonProcessingException _) {
          return encoded.lines().toList();
      }
  }

  String verifyPromptFor(String subtask, String result) {
      return """
              Adversarially verify the subagent result below: try to REFUTE it. Re-derive the \
              claims yourself with bash rather than trusting the result, and look for evidence \
              that contradicts them. Default to refuted if uncertain. Call report_findings with \
              summary 'refuted: <why>' or 'confirmed: <why>', citing the file:line or command \
              output that decided it.

              Subtask: %s

              Result to verify:
              %s""".formatted(subtask, result);
  }

  // invokeAll wartet auf jeden Job, daher ist jedes Future fertig, wenn es hier gelesen wird.
  List<String> runAll(ExecutorService pool, List<String> prompts, Model model) throws InterruptedException {
      List<Callable<String>> jobs = prompts.stream()
              .<Callable<String>>map(prompt -> () -> journaled(prompt, () -> runSubagent(model, prompt)))
              .toList();
      return pool.invokeAll(jobs).stream()
              .map(future -> switch (future.state()) {
                  case SUCCESS -> future.resultNow();
                  // Isolationsgrenze: Ein fehlerhafter Subagent soll nicht den ganzen Lauf beenden.
                  case FAILED -> "(subagent failed: " + future.exceptionNow() + ")";
                  case CANCELLED, RUNNING -> "(subagent failed: cancelled)";
              })
              .toList();
  }

  // Führe Teilaufgaben als parallele Subagenten aus, dann eine zweite Verifizierungswelle über
  // die Ergebnisse, und gib beides zurück. MAX_TOTAL_SUBTASKS begrenzt, wie viele das Modell
  // einreihen kann; MAX_CONCURRENT begrenzt, wie viele gleichzeitig laufen.
  ToolOutput runWorkflow(Model model, JsonValue rawSubtasks) throws InterruptedException {
      List<String> allSubtasks = normalizeSubtasks(rawSubtasks);
      List<String> subtasks = allSubtasks.stream().limit(MAX_TOTAL_SUBTASKS).toList();
      int dropped = allSubtasks.size() - subtasks.size();
      if (subtasks.isEmpty()) {
          return new ToolOutput("Workflow error: no usable subtasks were provided.", true);
      }
      System.err.println("[workflow] fanning out " + subtasks.size() + " agents");

      List<String> results;
      List<String> verdicts;
      try (ExecutorService pool = Executors.newFixedThreadPool(MAX_CONCURRENT, Thread.ofVirtual().factory())) {
          results = runAll(pool, subtasks, model);
          System.err.println("[workflow] verifying " + results.size() + " results");
          List<String> verifyPrompts = IntStream.range(0, subtasks.size())
                  .mapToObj(index -> verifyPromptFor(subtasks.get(index), results.get(index)))
                  .toList();
          verdicts = runAll(pool, verifyPrompts, model);
      }
      String joined = IntStream.range(0, subtasks.size())
              .mapToObj(index -> """
                      [agent %d: %s]
                      %s

                      [verify %d]
                      %s""".formatted(index + 1, subtasks.get(index), results.get(index), index + 1, verdicts.get(index)))
              .collect(Collectors.joining("\n\n"));
      if (dropped > 0) {
          joined = """
                  (note: %d subtasks beyond MAX_TOTAL_SUBTASKS=%d were not run; rerun them in a follow-up Workflow call)

                  %s""".formatted(dropped, MAX_TOTAL_SUBTASKS, joined);
      }
      return new ToolOutput(joined, false);
  }
  ```

  ```php PHP
  /**
   * Accept the subtasks input in whatever shape the model emits: an array, the array
   * JSON-encoded as a single string, or a newline-separated list.
   */
  function normalizeSubtasks(mixed $raw): array
  {
      if (is_string($raw)) {
          try {
              $raw = json_decode($raw, true, flags: JSON_THROW_ON_ERROR);
          } catch (\JsonException) {
              $raw = str_contains($raw, "\n") ? explode("\n", $raw) : [$raw];
          }
      }
      if (!is_array($raw)) {
          return [];
      }
      $tasks = array_map('trim', array_filter($raw, 'is_string'));
      return array_values(array_filter($tasks, fn ($task) => $task !== ''));
  }

  function verifyPromptFor(string $subtask, string $result): string
  {
      return 'Adversarially verify the subagent result below: try to REFUTE it. Re-derive the '
          . 'claims yourself with bash rather than trusting the result, and look for evidence '
          . 'that contradicts them. Default to refuted if uncertain. Call report_findings with '
          . "summary 'refuted: <why>' or 'confirmed: <why>', citing the file:line or command "
          . "output that decided it.\n\n"
          . <<<PROMPT
              Subtask: {$subtask}

              Result to verify:
              {$result}
              PROMPT;
  }

  /**
   * Run subtasks through the journal, then run a second verification wave over the
   * results, and return both. PHP's standard runtime has no lightweight thread pool,
   * so both waves run sequentially here (MAX_CONCURRENT is unused); the SDK examples
   * in other languages fan them out in parallel.
   */
  function runWorkflow(Client $client, string $model, mixed $rawSubtasks): array
  {
      $allSubtasks = normalizeSubtasks($rawSubtasks);
      $subtasks = array_slice($allSubtasks, 0, MAX_TOTAL_SUBTASKS);
      $dropped = count($allSubtasks) - count($subtasks);
      if ($subtasks === []) {
          return ['Workflow error: no usable subtasks were provided.', true];
      }
      fwrite(STDERR, '[workflow] running ' . count($subtasks) . " agents\n");

      $runOne = function (string $prompt) use ($client, $model): string {
          try {
              return journaled($prompt, fn () => runSubagent($client, $model, $prompt));
          } catch (\Throwable $error) {
              // Isolationsgrenze: ein fehlerhafter Subagent soll nicht den gesamten Lauf beenden.
              return '(subagent failed: ' . $error::class . ': ' . $error->getMessage() . ')';
          }
      };

      $results = array_map($runOne, $subtasks);
      fwrite(STDERR, '[workflow] verifying ' . count($results) . " results\n");
      $verifyPrompts = array_map(verifyPromptFor(...), $subtasks, $results);
      $verdicts = array_map($runOne, $verifyPrompts);

      $sections = [];
      foreach ($subtasks as $index => $task) {
          $sections[] = '[agent ' . ($index + 1) . ": {$task}]\n{$results[$index]}"
              . "\n\n[verify " . ($index + 1) . "]\n{$verdicts[$index]}";
      }
      $joined = implode("\n\n", $sections);
      if ($dropped > 0) {
          $joined = '(note: ' . $dropped . ' subtasks beyond MAX_TOTAL_SUBTASKS=' . MAX_TOTAL_SUBTASKS
              . " were not run; rerun them in a follow-up Workflow call)\n\n" . $joined;
      }
      return [$joined, false];
  }
  ```

  ```ruby Ruby
  # Akzeptiere den subtasks-Input in jeder Form, die das Modell ausgibt: ein Array, das Array
  # als einzelner JSON-kodierter String oder eine durch Zeilenumbrüche getrennte Liste.
  def normalize_subtasks(raw)
    if raw.is_a?(String)
      begin
        raw = JSON.parse(raw)
      rescue JSON::ParserError
        raw = raw.include?("\n") ? raw.split("\n") : [raw]
      end
    end
    return [] unless raw.is_a?(Array)
    raw.select { |task| task.is_a?(String) }.map(&:strip).reject(&:empty?)
  end

  def verify_prompt_for(subtask, result)
    "Adversarially verify the subagent result below: try to REFUTE it. Re-derive the " \
      "claims yourself with bash rather than trusting the result, and look for evidence " \
      "that contradicts them. Default to refuted if uncertain. Call report_findings with " \
      "summary 'refuted: <why>' or 'confirmed: <why>', citing the file:line or command " \
      "output that decided it.\n\n" +
      <<~PROMPT.chomp
        Subtask: #{subtask}

        Result to verify:
        #{result}
      PROMPT
  end

  # Map mit Nebenläufigkeitslimit: höchstens `limit` Threads laufen gleichzeitig.
  def map_with_limit(items, limit)
    results = Array.new(items.length)
    queue = Queue.new
    items.each_with_index { |item, index| queue << [index, item] }
    workers = Array.new([limit, items.length].min) do
      Thread.new do
        until queue.empty?
          index, item = queue.pop(true) rescue break
          results[index] = yield item
        end
      end
    end
    workers.each(&:join)
    results
  end

  # Führe Teilaufgaben als parallele Subagenten aus, lasse dann eine zweite Verifikationswelle über
  # die Ergebnisse laufen und gib beides zurück. MAX_TOTAL_SUBTASKS begrenzt, wie viele das Modell
  # einreihen kann; MAX_CONCURRENT begrenzt, wie viele gleichzeitig laufen.
  def run_workflow(model, raw_subtasks)
    all_subtasks = normalize_subtasks(raw_subtasks)
    subtasks = all_subtasks.first(MAX_TOTAL_SUBTASKS)
    dropped = all_subtasks.length - subtasks.length
    return ["Workflow error: no usable subtasks were provided.", true] if subtasks.empty?

    warn "[workflow] fanning out #{subtasks.length} agents"
    run_one = lambda do |prompt|
      journaled(prompt) { run_subagent(model, prompt) }
    rescue => error # isolation boundary: one bad subagent should not end the run
      "(subagent failed: #{error.class}: #{error.message})"
    end

    results = map_with_limit(subtasks, MAX_CONCURRENT, &run_one)
    warn "[workflow] verifying #{results.length} results"
    verify_prompts = subtasks.zip(results).map { |task, result| verify_prompt_for(task, result) }
    verdicts = map_with_limit(verify_prompts, MAX_CONCURRENT, &run_one)

    joined = subtasks.each_with_index.map do |task, index|
      <<~ENTRY.chomp
        [agent #{index + 1}: #{task}]
        #{results[index]}

        [verify #{index + 1}]
        #{verdicts[index]}
      ENTRY
    end.join("\n\n")
    if dropped > 0
      joined =
        "(note: #{dropped} subtasks beyond MAX_TOTAL_SUBTASKS=#{MAX_TOTAL_SUBTASKS} were not " \
        "run; rerun them in a follow-up Workflow call)\n\n#{joined}"
    end
    [joined, false]
  end
  ```
</CodeGroup>

## Den Modus mit System-Nachrichten mitten im Gespräch umschalten

Der Agent hängt zuerst die Nachricht des Nutzers an, dann alle fälligen System-Nachrichten: den Austrittshinweis, den vollständigen Modus-Text beim Eintritt oder die periodische Auffrischung. Die System-Nachricht nach dem Nutzer-Turn zu platzieren, lässt jedes gecachte Byte davor unangetastet und erfüllt die Platzierungsregel, dass eine System-Nachricht auf einen Nutzer-Turn folgt.

<CodeGroup>
  ```bash cURL
  # Ein Turn im Orchestrierungsmodus: der Modus-Reminder steckt im messages-Array als
  # {"role": "system"}-Eintrag nach dem User-Turn, für den er gilt. Die Antwort
  # stoppt beim ersten Tool-Aufruf. Der Agent-Loop, der Tool-Aufrufe ausführt und
  # Subagenten verteilt, wird in den SDK-Tabs gezeigt; die Workflow-Beschreibung ist hier
  # gekürzt, die SDK-Beispiele enthalten den vollständigen Standing-Consent-Text.
  curl --fail-with-body -sS https://api.anthropic.com/v1/messages \
    -H "x-api-key: $ANTHROPIC_API_KEY" \
    -H "anthropic-version: 2023-06-01" \
    -H "content-type: application/json" \
    -d '{
      "model": "claude-opus-5-5",
      "max_tokens": 64000,
      "system": "You are a helpful general-purpose agent. Answer the request from the user directly.",
      "output_config": {"effort": "xhigh"},
      "tools": [
        {
          "name": "Workflow",
          "description": "Orchestrate a multiagent workflow: split a large task into independent subtasks and run them as parallel agents, then collect their results. Opt-in: only use this tool when the user explicitly asks for a workflow, or when a system message confirms that orchestration mode is on. Granularity: scope each subtask to a distinct concern rather than per line or file section; a review of a small module rarely needs more than about ten. Standing consent: while a system message confirms orchestration mode is on, author and run a workflow for every substantive task by default; work solo only on conversational turns or trivial mechanical edits.",
          "input_schema": {
            "type": "object",
            "properties": {
              "subtasks": {
                "type": "array",
                "items": {"type": "string"},
                "description": "Independent subtask prompts to run as parallel agents"
              }
            },
            "required": ["subtasks"]
          }
        },
        {"type": "bash_20250124", "name": "bash"}
      ],
      "messages": [
        {
          "role": "user",
          "content": "Explore the current directory, then give a thorough review: what it does, code-quality issues, and concrete improvements."
        },
        {
          "role": "system",
          "content": "Orchestration mode is on: optimize for the most exhaustive, correct answer rather than the fastest one. Use the Workflow tool on every substantive task, sized to the natural decomposition of the problem rather than the maximum the tool allows. See the description of the Workflow tool for standing consent, granularity guidance, and quality patterns. Work solo only on conversational or trivial turns."
        }
      ]
    }'
  ```

  ```bash CLI
  # Ein Orchestrierungsmodus-Turn: Der Modus-Reminder steckt im messages-Array als
  # system-Rollen-Eintrag nach dem User-Turn, auf den er sich bezieht. Die Antwort stoppt
  # beim ersten Tool-Aufruf. Der Agenten-Loop, der Tool-Aufrufe ausführt und
  # Subagenten auffächert, wird in den SDK-Tabs gezeigt; die Workflow-Beschreibung ist hier gekürzt,
  # die SDK-Beispiele enthalten den vollständigen Standing-Consent-Text.
  ant messages create <<'YAML'
  model: claude-opus-5-5
  max_tokens: 64000
  system: You are a helpful general-purpose agent. Answer the user's request directly.
  output_config: {effort: xhigh}
  tools:
    - name: Workflow
      description: >-
        Orchestrate a multiagent workflow: split a large task into independent
        subtasks and run them as parallel agents, then collect their results.
        Opt-in: only use this tool when the user explicitly asks for a workflow,
        or when a system message confirms that orchestration mode is on.
        Granularity: scope each subtask to a distinct concern rather than per
        line or file section; a review of a small module rarely needs more than
        about ten. Standing consent: while a system message confirms
        orchestration mode is on, author and run a workflow for every
        substantive task by default; work solo only on conversational turns or
        trivial mechanical edits.
      input_schema:
        type: object
        properties:
          subtasks:
            type: array
            items: {type: string}
            description: Independent subtask prompts to run as parallel agents
        required: [subtasks]
    - {type: bash_20250124, name: bash}
  messages:
    - role: user
      content: >-
        Explore the current directory, then give a thorough review: what it
        does, code-quality issues, and concrete improvements.
    - role: system
      content: >-
        Orchestration mode is on: optimize for the most exhaustive, correct
        answer rather than the fastest one. Use the Workflow tool on every
        substantive task, sized to the problem's natural decomposition rather
        than the maximum the tool allows. See the Workflow tool's description
        for standing consent, granularity guidance, and quality patterns. Work
        solo only on conversational or trivial turns.
  YAML
  ```

  ```python Python
  class ModeAgent:
      """An agent loop whose orchestration mode is toggled with mid-conversation system messages."""

      def __init__(self, model: str, mode_on: bool = True):
          self.model = model
          self.mode_on = mode_on
          self.messages: list[dict] = []
          self._mode_announced = False
          self._exit_pending = False
          self._turns_since_reminder = 0

      def set_mode(self, mode_on: bool) -> None:
          """Turn the mode on or off. The notice is delivered with the next user turn."""
          if mode_on == self.mode_on:
              return
          if not mode_on:
              if self._mode_announced:
                  self._exit_pending = True
          else:
              self._exit_pending = False
          self.mode_on = mode_on

      def _due_system_messages(self) -> list[dict]:
          """System messages owed on this turn: an exit notice, the full mode text on entry,
          or a one-line refresher every TURNS_BETWEEN_REFRESHERS user turns."""
          due = []
          if self._exit_pending:
              self._exit_pending = False
              self._mode_announced = False
              due.append({"role": "system", "content": MODE_EXIT})
          if self.mode_on:
              if not self._mode_announced:
                  self._mode_announced = True
                  self._turns_since_reminder = 0
                  due.append({"role": "system", "content": MODE_ENTER})
              elif self._turns_since_reminder >= TURNS_BETWEEN_REFRESHERS:
                  self._turns_since_reminder = 0
                  due.append({"role": "system", "content": MODE_REFRESH})
          return due

      def turn(self, user_input: str) -> str:
          # System-Nachrichten mitten in der Konversation folgen auf den User-Turn, für den sie gelten, wodurch
          # der gecachte Präfix davor unangetastet bleibt.
          self.messages.append({"role": "user", "content": user_input})
          self.messages.extend(self._due_system_messages())
          self._turns_since_reminder += 1

          for _ in range(MAX_MAIN_TURNS):
              with client.messages.stream(
                  model=self.model,
                  max_tokens=64000,
                  system=SYSTEM_PROMPT,  # static for the whole session
                  output_config={"effort": EFFORT},
                  tools=[WORKFLOW_TOOL, BASH_TOOL],
                  messages=self.messages,
                  timeout=REQUEST_TIMEOUT_SECONDS,
              ) as stream:
                  response = stream.get_final_message()
              self.messages.append({"role": "assistant", "content": response.content})

              if response.stop_reason == "pause_turn":
                  continue
              if response.stop_reason != "tool_use":
                  text = "".join(
                      block.text for block in response.content if block.type == "text"
                  )
                  if response.stop_reason == "max_tokens":
                      # Verwirf die abgeschnittene Assistant-Nachricht, damit spätere Turns nicht darauf aufbauen.
                      self.messages.pop()
                      text += "\n\n(warning: response was truncated at max_tokens)"
                  return text

              tool_results = []
              for block in response.content:
                  if block.type != "tool_use":
                      continue
                  match block.name:
                      case "Workflow":
                          output, is_error = run_workflow(
                              self.model, block.input.get("subtasks", [])
                          )
                      case "bash":
                          output, is_error = handle_bash_block(block)
                      case _:
                          output, is_error = f"unknown tool: {block.name}", True
                  tool_results.append(
                      {
                          "type": "tool_result",
                          "tool_use_id": block.id,
                          "content": output,
                          "is_error": is_error,
                      }
                  )
              self.messages.append({"role": "user", "content": tool_results})
          return "(hit the main loop turn limit before finishing)"
  ```

  ```typescript TypeScript
  // Eine Agenten-Schleife, deren Orchestrierungsmodus per System-Nachrichten mitten im Gespräch umgeschaltet wird.
  class ModeAgent {
    readonly #model: string;
    readonly #messages: Anthropic.MessageParam[] = [];
    #modeOn: boolean;
    #modeAnnounced = false;
    #exitPending = false;
    #turnsSinceReminder = 0;

    constructor(model: string, modeOn = true) {
      this.#model = model;
      this.#modeOn = modeOn;
    }

    // Schaltet den Modus ein oder aus. Der Hinweis wird mit dem nächsten User-Turn zugestellt.
    setMode(modeOn: boolean): void {
      if (modeOn === this.#modeOn) {
        return;
      }
      // Ein Beendigungshinweis ist nur nötig, wenn dem Modell gesagt wurde, dass der Modus an war.
      this.#exitPending = !modeOn && this.#modeAnnounced;
      this.#modeOn = modeOn;
    }

    // In diesem Turn fällige System-Nachrichten: ein Hinweis beim Beenden, der volle Modustext beim Einstieg,
    // oder eine einzeilige Auffrischung alle TURNS_BETWEEN_REFRESHERS User-Turns.
    #dueSystemMessages(): Anthropic.MessageParam[] {
      const due: Anthropic.MessageParam[] = [];
      if (this.#exitPending) {
        this.#exitPending = false;
        this.#modeAnnounced = false;
        due.push({ role: "system", content: MODE_EXIT });
      }
      if (this.#modeOn) {
        if (!this.#modeAnnounced) {
          this.#modeAnnounced = true;
          this.#turnsSinceReminder = 0;
          due.push({ role: "system", content: MODE_ENTER });
        } else if (this.#turnsSinceReminder >= TURNS_BETWEEN_REFRESHERS) {
          this.#turnsSinceReminder = 0;
          due.push({ role: "system", content: MODE_REFRESH });
        }
      }
      return due;
    }

    async turn(userInput: string): Promise<string> {
      // System-Nachrichten mitten im Gespräch folgen auf den User-Turn, für den sie gelten; so bleibt
      // das gecachte Präfix davor unangetastet.
      this.#messages.push({ role: "user", content: userInput }, ...this.#dueSystemMessages());
      this.#turnsSinceReminder += 1;

      for (let turn = 0; turn < MAX_MAIN_TURNS; turn++) {
        const response = await client.messages
          .stream(
            {
              model: this.#model,
              max_tokens: 64000,
              system: SYSTEM_PROMPT, // static for the whole session
              output_config: { effort: EFFORT },
              tools: [WORKFLOW_TOOL, BASH_TOOL],
              messages: this.#messages
            },
            { signal: AbortSignal.timeout(REQUEST_TIMEOUT_SECONDS * 1000) }
          )
          .finalMessage();
        this.#messages.push({ role: "assistant", content: response.content });

        if (response.stop_reason === "pause_turn") {
          continue;
        }
        if (response.stop_reason !== "tool_use") {
          let text = response.content
            .flatMap((block) => (block.type === "text" ? [block.text] : []))
            .join("");
          if (response.stop_reason === "max_tokens") {
            // Verwirft die abgeschnittene Assistant-Nachricht, damit spätere Turns nicht darauf aufbauen.
            this.#messages.pop();
            text += "\n\n(warning: response was truncated at max_tokens)";
          }
          return text;
        }

        const toolResults: Anthropic.ToolResultBlockParam[] = [];
        for (const block of response.content) {
          if (block.type !== "tool_use") {
            continue;
          }
          let outcome: ToolOutcome;
          switch (block.name) {
            case "Workflow": {
              const { subtasks } = block.input as { subtasks?: unknown };
              outcome = await runWorkflow(this.#model, subtasks);
              break;
            }
            case "bash":
              outcome = await handleBashBlock(block);
              break;
            default:
              outcome = { output: `unknown tool: ${block.name}`, isError: true };
          }
          toolResults.push({
            type: "tool_result",
            tool_use_id: block.id,
            content: outcome.output,
            is_error: outcome.isError
          });
        }
        this.#messages.push({ role: "user", content: toolResults });
      }
      return "(hit the main loop turn limit before finishing)";
    }
  }
  ```

  ```csharp C#
  // Eine Agenten-Schleife, deren Orchestrierungsmodus per System-Nachrichten mitten im Gespräch umgeschaltet wird.
  List<MessageParam> messages = [];
  var modeOn = true;
  var modeAnnounced = false;
  var exitPending = false;
  var turnsSinceReminder = 0;

  // Schalte den Modus ein oder aus. Die Notiz wird mit dem nächsten User-Turn zugestellt.
  void SetMode(bool nextModeOn)
  {
      if (nextModeOn == modeOn)
      {
          return;
      }
      // Ein Beendigungshinweis ist nur fällig, wenn dem Modell gesagt wurde, dass der Modus an war.
      exitPending = !nextModeOn && modeAnnounced;
      modeOn = nextModeOn;
  }

  MessageParam SystemMessage(string content) => new() { Role = Role.System, Content = content };

  // Auf diesem Turn fällige System-Nachrichten: eine Beendigungsnotiz, der volle Modus-Text beim Eintritt
  // oder eine einzeilige Auffrischung alle turnsBetweenRefreshers User-Turns.
  List<MessageParam> DueSystemMessages()
  {
      List<MessageParam> due = [];
      if (exitPending)
      {
          exitPending = false;
          modeAnnounced = false;
          due.Add(SystemMessage(modeExit));
      }
      if (modeOn)
      {
          if (!modeAnnounced)
          {
              modeAnnounced = true;
              turnsSinceReminder = 0;
              due.Add(SystemMessage(modeEnter));
          }
          else if (turnsSinceReminder >= turnsBetweenRefreshers)
          {
              turnsSinceReminder = 0;
              due.Add(SystemMessage(modeRefresh));
          }
      }
      return due;
  }

  // Schicke einen User-Turn durch die Schleife und führe Tool-Aufrufe aus, bis das Modell stoppt.
  async Task<string> Turn(string userInput)
  {
      // System-Nachrichten mitten im Gespräch folgen dem User-Turn, für den sie gelten, wodurch
      // das gecachte Präfix davor unangetastet bleibt.
      messages.Add(new() { Role = Role.User, Content = userInput });
      messages.AddRange(DueSystemMessages());
      turnsSinceReminder++;

      for (var turn = 0; turn < maxMainTurns; turn++)
      {
          using var deadline = new CancellationTokenSource(TimeSpan.FromSeconds(requestTimeoutSeconds));
          var response = await client.Messages.CreateStreaming(new MessageCreateParams
          {
              Model = model,
              MaxTokens = 64000,
              System = systemPrompt, // static for the whole session
              OutputConfig = new OutputConfig { Effort = effort },
              Tools = [workflowTool, bashTool],
              Messages = messages,
          }, cancellationToken: deadline.Token).Aggregate();
          messages.Add(new()
          {
              Role = Role.Assistant,
              Content = response.Content.Select(block => new ContentBlockParam(block.Json)).ToList(),
          });

          if (response.StopReason == StopReason.PauseTurn)
          {
              continue;
          }
          if (response.StopReason != StopReason.ToolUse)
          {
              var text = string.Concat(
                  response.Content.Select(block => block.TryPickText(out var textBlock) ? textBlock.Text : ""));
              if (response.StopReason == StopReason.MaxTokens)
              {
                  // Verwirf die abgeschnittene Assistant-Nachricht, damit spätere Turns nicht darauf aufbauen.
                  messages.RemoveAt(messages.Count - 1);
                  text += "\n\n(warning: response was truncated at max_tokens)";
              }
              return text;
          }

          List<ContentBlockParam> toolResults = [];
          foreach (var block in response.Content)
          {
              if (!block.TryPickToolUse(out var toolUse))
              {
                  continue;
              }
              var (output, isError) = toolUse.Name switch
              {
                  "Workflow" => await RunWorkflow(toolUse.Input.GetValueOrDefault("subtasks")),
                  "bash" => await HandleBashBlock(toolUse),
                  _ => ($"unknown tool: {toolUse.Name}", true),
              };
              toolResults.Add(new ToolResultBlockParam(toolUse.ID) { Content = output, IsError = isError });
          }
          messages.Add(new() { Role = Role.User, Content = toolResults });
      }
      return "(hit the main loop turn limit before finishing)";
  }
  ```

  ```go Go
  // modeAgent ist eine Agenten-Schleife, deren Orchestrierungsmodus per System-Nachrichten
  // mitten im Gespräch umgeschaltet wird.
  type modeAgent struct {
  	model              string
  	modeOn             bool
  	messages           []anthropic.MessageParam
  	modeAnnounced      bool
  	exitPending        bool
  	turnsSinceReminder int
  }

  func newModeAgent(model string) *modeAgent {
  	return &modeAgent{model: model, modeOn: true}
  }

  // setMode schaltet den Modus ein oder aus. Die Notiz wird mit dem nächsten User-Turn zugestellt.
  func (agent *modeAgent) setMode(modeOn bool) {
  	if modeOn == agent.modeOn {
  		return
  	}
  	if !modeOn {
  		if agent.modeAnnounced {
  			agent.exitPending = true
  		}
  	} else {
  		agent.exitPending = false
  	}
  	agent.modeOn = modeOn
  }

  // dueSystemMessages gibt die für diesen Turn fälligen System-Nachrichten zurück: eine Beendigungsnotiz,
  // den vollen Modus-Text beim Eintritt oder einen Einzeiler alle turnsBetweenRefreshers User-Turns.
  func (agent *modeAgent) dueSystemMessages() []anthropic.MessageParam {
  	systemMessage := func(content string) anthropic.MessageParam {
  		return anthropic.MessageParam{
  			Role:    anthropic.MessageParamRoleSystem,
  			Content: []anthropic.ContentBlockParamUnion{anthropic.NewTextBlock(content)},
  		}
  	}
  	var due []anthropic.MessageParam
  	if agent.exitPending {
  		agent.exitPending = false
  		agent.modeAnnounced = false
  		due = append(due, systemMessage(modeExit))
  	}
  	if agent.modeOn {
  		if !agent.modeAnnounced {
  			agent.modeAnnounced = true
  			agent.turnsSinceReminder = 0
  			due = append(due, systemMessage(modeEnter))
  		} else if agent.turnsSinceReminder >= turnsBetweenRefreshers {
  			agent.turnsSinceReminder = 0
  			due = append(due, systemMessage(modeRefresh))
  		}
  	}
  	return due
  }

  // turn schickt einen User-Turn durch die Schleife und führt Tool-Aufrufe aus, bis das Modell stoppt.
  func (agent *modeAgent) turn(ctx context.Context, userInput string) (string, error) {
  	// System-Nachrichten mitten im Gespräch folgen dem User-Turn, für den sie gelten, wodurch
  	// das gecachte Präfix davor unangetastet bleibt.
  	agent.messages = append(agent.messages, anthropic.NewUserMessage(anthropic.NewTextBlock(userInput)))
  	agent.messages = append(agent.messages, agent.dueSystemMessages()...)
  	agent.turnsSinceReminder++

  	for range maxMainTurns {
  		response, err := streamMessage(ctx, anthropic.MessageNewParams{
  			Model:        agent.model,
  			MaxTokens:    64000,
  			System:       []anthropic.TextBlockParam{{Text: systemPrompt}}, // static for the whole session
  			OutputConfig: anthropic.OutputConfigParam{Effort: effort},
  			Tools:        []anthropic.ToolUnionParam{workflowTool, bashTool},
  			Messages:     agent.messages,
  		})
  		if err != nil {
  			return "", err
  		}
  		agent.messages = append(agent.messages, response.ToParam())

  		switch response.StopReason {
  		case anthropic.StopReasonPauseTurn:
  			continue
  		case anthropic.StopReasonMaxTokens:
  			// Verwirf die abgeschnittene Assistant-Nachricht, damit spätere Turns nicht darauf aufbauen.
  			agent.messages = agent.messages[:len(agent.messages)-1]
  			return textOf(response) + "\n\n(warning: response was truncated at max_tokens)", nil
  		case anthropic.StopReasonToolUse:
  		default:
  			return textOf(response), nil
  		}

  		var toolResults []anthropic.ContentBlockParamUnion
  		for _, block := range response.Content {
  			toolUse, ok := block.AsAny().(anthropic.ToolUseBlock)
  			if !ok {
  				continue
  			}
  			var output string
  			var isError bool
  			switch toolUse.Name {
  			case "Workflow":
  				var input struct {
  					Subtasks json.RawMessage `json:"subtasks"`
  				}
  				if err := json.Unmarshal(toolUse.Input, &input); err != nil {
  					output, isError = fmt.Sprintf("Workflow error: could not parse input: %s", err), true
  				} else {
  					output, isError = runWorkflow(ctx, agent.model, input.Subtasks)
  				}
  			case "bash":
  				output, isError = handleBashBlock(ctx, toolUse)
  			default:
  				output, isError = "unknown tool: "+toolUse.Name, true
  			}
  			toolResults = append(toolResults, anthropic.NewToolResultBlock(toolUse.ID, output, isError))
  		}
  		agent.messages = append(agent.messages, anthropic.NewUserMessage(toolResults...))
  	}
  	return "(hit the main loop turn limit before finishing)", nil
  }
  ```

  ```java Java
  // Eine Agenten-Schleife, deren Orchestrierungsmodus per System-Nachrichten mitten im Gespräch umgeschaltet wird.
  class ModeAgent {
      private final Model model;
      private final List<MessageParam> messages = new ArrayList<>();
      private boolean modeOn;
      private boolean modeAnnounced;
      private boolean exitPending;
      private int turnsSinceReminder;

      ModeAgent(Model model) {
          this(model, true);
      }

      ModeAgent(Model model, boolean modeOn) {
          this.model = model;
          this.modeOn = modeOn;
      }

      // Schalte den Modus ein oder aus. Der Hinweis wird mit dem nächsten User-Turn zugestellt:
      // ein Beenden-Hinweis ist nur nötig, wenn dem Modell gesagt wurde, dass der Modus an war.
      void setMode(boolean modeOn) {
          if (modeOn == this.modeOn) {
              return;
          }
          exitPending = !modeOn && modeAnnounced;
          this.modeOn = modeOn;
      }

      // In diesem Turn fällige System-Nachrichten: eine Beendigungsnotiz, der vollständige Modus-Text beim Eintritt
      // oder eine einzeilige Auffrischung alle TURNS_BETWEEN_REFRESHERS User-Turns.
      private List<MessageParam> dueSystemMessages() {
          List<MessageParam> due = new ArrayList<>();
          if (exitPending) {
              exitPending = false;
              modeAnnounced = false;
              due.add(systemMessage(MODE_EXIT));
          }
          if (modeOn) {
              if (!modeAnnounced) {
                  modeAnnounced = true;
                  turnsSinceReminder = 0;
                  due.add(systemMessage(MODE_ENTER));
              } else if (turnsSinceReminder >= TURNS_BETWEEN_REFRESHERS) {
                  turnsSinceReminder = 0;
                  due.add(systemMessage(MODE_REFRESH));
              }
          }
          return due;
      }

      private MessageParam systemMessage(String content) {
          return MessageParam.builder()
                  .role(MessageParam.Role.SYSTEM)
                  .content(content)
                  .build();
      }

      // Sende einen User-Turn durch die Schleife und führe Tool-Aufrufe aus, bis das Modell stoppt.
      String turn(String userInput) throws InterruptedException {
          // System-Nachrichten mitten im Gespräch folgen dem User-Turn, für den sie gelten, wodurch
          // das gecachte Präfix davor unberührt bleibt.
          messages.add(MessageParam.builder().role(MessageParam.Role.USER).content(userInput).build());
          messages.addAll(dueSystemMessages());
          turnsSinceReminder++;

          for (int turn = 0; turn < MAX_MAIN_TURNS; turn++) {
              MessageCreateParams params = MessageCreateParams.builder()
                      .model(model)
                      .maxTokens(64000L)
                      .system(SYSTEM_PROMPT) // static for the whole session
                      .outputConfig(OutputConfig.builder().effort(EFFORT).build())
                      .addTool(WORKFLOW_TOOL)
                      .addTool(BASH_TOOL)
                      .messages(messages)
                      .build();
              MessageAccumulator accumulator = MessageAccumulator.create();
              try (var stream = client.messages().createStreaming(params, REQUEST_OPTIONS)) {
                  stream.stream().forEach(accumulator::accumulate);
              }
              Message response = accumulator.message();
              messages.add(response.toParam());

              StopReason stopReason = response.stopReason().orElse(null);
              if (StopReason.PAUSE_TURN.equals(stopReason)) {
                  continue;
              }
              if (!StopReason.TOOL_USE.equals(stopReason)) {
                  String text = response.content().stream()
                          .flatMap(block -> block.text().stream())
                          .map(TextBlock::text)
                          .collect(Collectors.joining());
                  if (StopReason.MAX_TOKENS.equals(stopReason)) {
                      // Verwirf die abgeschnittene Assistant-Nachricht, damit spätere Turns nicht darauf aufbauen.
                      messages.removeLast();
                      text += "\n\n(warning: response was truncated at max_tokens)";
                  }
                  return text;
              }

              List<ToolUseBlock> toolUses = response.content().stream()
                      .flatMap(block -> block.toolUse().stream())
                      .toList();
              List<ContentBlockParam> toolResults = new ArrayList<>();
              for (ToolUseBlock toolUse : toolUses) {
                  ToolOutput result = switch (toolUse.name()) {
                      case "Workflow" -> runWorkflow(model,
                              toolInput(toolUse).getOrDefault("subtasks", JsonValue.from(List.of())));
                      case "bash" -> handleBashBlock(toolUse);
                      default -> new ToolOutput("unknown tool: " + toolUse.name(), true);
                  };
                  toolResults.add(ContentBlockParam.ofToolResult(ToolResultBlockParam.builder()
                          .toolUseId(toolUse.id())
                          .content(result.output())
                          .isError(result.isError())
                          .build()));
              }
              messages.add(MessageParam.builder()
                      .role(MessageParam.Role.USER)
                      .contentOfBlockParams(toolResults)
                      .build());
          }
          return "(hit the main loop turn limit before finishing)";
      }
  }
  ```

  ```php PHP
  /** An agent loop whose orchestration mode is toggled with mid-conversation system messages. */
  class ModeAgent
  {
      private array $messages = [];
      private bool $modeAnnounced = false;
      private bool $exitPending = false;
      private int $turnsSinceReminder = 0;

      public function __construct(
          private readonly Client $client,
          private readonly string $model,
          private bool $modeOn = true,
      ) {
      }

      /** Turn the mode on or off. The notice is delivered with the next user turn. */
      public function setMode(bool $modeOn): void
      {
          if ($modeOn === $this->modeOn) {
              return;
          }
          if ($modeOn) {
              $this->exitPending = false;
          } elseif ($this->modeAnnounced) {
              $this->exitPending = true;
          }
          $this->modeOn = $modeOn;
      }

      public function turn(string $userInput): string
      {
          // System-Messages mitten im Gespräch folgen auf den User-Turn, für den sie gelten; so bleibt
          // der gecachte Präfix davor unangetastet.
          $this->messages[] = ['role' => 'user', 'content' => $userInput];
          array_push($this->messages, ...$this->dueSystemMessages());
          $this->turnsSinceReminder++;

          for ($turn = 0; $turn < MAX_MAIN_TURNS; $turn++) {
              $stream = $this->client->messages->createStream(
                  model: $this->model,
                  maxTokens: 64000,
                  system: SYSTEM_PROMPT, // static for the whole session
                  outputConfig: ['effort' => EFFORT],
                  tools: [WORKFLOW_TOOL, BASH_TOOL],
                  messages: $this->messages,
                  requestOptions: ['timeout' => REQUEST_TIMEOUT_SECONDS],
              );
              [$content, $stopReason] = drainMessageStream($stream);
              $this->messages[] = ['role' => 'assistant', 'content' => $content];

              if ($stopReason === 'pause_turn') {
                  continue;
              }
              if ($stopReason !== 'tool_use') {
                  $text = '';
                  foreach ($content as $block) {
                      if ($block instanceof \Anthropic\Messages\TextBlock) {
                          $text .= $block->text;
                      }
                  }
                  if ($stopReason === 'max_tokens') {
                      // Verwirf die abgeschnittene Assistant-Nachricht, damit spätere Turns nicht darauf aufbauen.
                      array_pop($this->messages);
                      $text .= "\n\n(warning: response was truncated at max_tokens)";
                  }
                  return $text;
              }

              $toolResults = [];
              foreach ($content as $block) {
                  if (!$block instanceof \Anthropic\Messages\ToolUseBlock) {
                      continue;
                  }
                  if ($block->name === 'Workflow') {
                      [$output, $isError] =
                          runWorkflow($this->client, $this->model, $block->input['subtasks'] ?? []);
                  } elseif ($block->name === 'bash') {
                      [$output, $isError] = handleBashBlock($block);
                  } else {
                      $output = "unknown tool: {$block->name}";
                      $isError = true;
                  }
                  $toolResults[] = [
                      'type' => 'tool_result',
                      'tool_use_id' => $block->id,
                      'content' => $output,
                      'is_error' => $isError,
                  ];
              }
              $this->messages[] = ['role' => 'user', 'content' => $toolResults];
          }
          return '(hit the main loop turn limit before finishing)';
      }

      /**
       * System messages owed on this turn: an exit notice, the full mode text on entry,
       * or a one-line refresher every TURNS_BETWEEN_REFRESHERS user turns.
       */
      private function dueSystemMessages(): array
      {
          $due = [];
          if ($this->exitPending) {
              $this->exitPending = false;
              $this->modeAnnounced = false;
              $due[] = ['role' => 'system', 'content' => MODE_EXIT];
          }
          if ($this->modeOn) {
              if (!$this->modeAnnounced) {
                  $this->modeAnnounced = true;
                  $this->turnsSinceReminder = 0;
                  $due[] = ['role' => 'system', 'content' => MODE_ENTER];
              } elseif ($this->turnsSinceReminder >= TURNS_BETWEEN_REFRESHERS) {
                  $this->turnsSinceReminder = 0;
                  $due[] = ['role' => 'system', 'content' => MODE_REFRESH];
              }
          }
          return $due;
      }
  }
  ```

  ```ruby Ruby
  # Eine Agenten-Schleife, deren Orchestrierungsmodus per System-Messages mitten im Gespräch umgeschaltet wird.
  class ModeAgent
    def initialize(model, mode_on: true)
      @model = model
      @mode_on = mode_on
      @messages = []
      @mode_announced = false
      @exit_pending = false
      @turns_since_reminder = 0
    end

    # Schalte den Modus ein oder aus. Der Hinweis wird mit dem nächsten User-Turn übermittelt.
    def set_mode(mode_on)
      return if mode_on == @mode_on

      if mode_on
        @exit_pending = false
      else
        @exit_pending = true if @mode_announced
      end
      @mode_on = mode_on
    end

    def turn(user_input)
      # System-Messages mitten im Gespräch folgen auf den User-Turn, für den sie gelten; so bleibt
      # der gecachte Präfix davor unangetastet.
      @messages << {role: "user", content: user_input}
      @messages.concat(due_system_messages)
      @turns_since_reminder += 1

      MAX_MAIN_TURNS.times do
        stream = CLIENT.messages.stream(
          model: @model,
          max_tokens: 64_000,
          system_: SYSTEM_PROMPT, # static for the whole session
          output_config: {effort: EFFORT},
          tools: [WORKFLOW_TOOL, BASH_TOOL],
          messages: @messages,
          request_options: {timeout: REQUEST_TIMEOUT_SECONDS}
        )
        response = stream.accumulated_message
        @messages << {role: "assistant", content: assistant_content_param(response.content)}

        next if response.stop_reason == :pause_turn

        unless response.stop_reason == :tool_use
          text = response.content.select { |block| block.type == :text }.map(&:text).join
          if response.stop_reason == :max_tokens
            # Verwirf die abgeschnittene Assistant-Nachricht, damit spätere Turns nicht darauf aufbauen.
            @messages.pop
            text += "\n\n(warning: response was truncated at max_tokens)"
          end
          return text
        end

        tool_results = []
        response.content.each do |block|
          next unless block.type == :tool_use

          input = parse_tool_input(block.input)
          case block.name
          when "Workflow"
            output, is_error = run_workflow(@model, input["subtasks"] || [])
          when "bash"
            output, is_error = handle_bash_block(block)
          else
            output, is_error = "unknown tool: #{block.name}", true
          end
          tool_results << {
            type: "tool_result",
            tool_use_id: block.id,
            content: output,
            is_error: is_error
          }
        end
        @messages << {role: "user", content: tool_results}
      end
      "(hit the main loop turn limit before finishing)"
    end

    private

    # In diesem Turn fällige System-Messages: ein Hinweis beim Verlassen, der volle Modus-Text beim Einstieg
    # oder eine einzeilige Auffrischung alle TURNS_BETWEEN_REFRESHERS User-Turns.
    def due_system_messages
      due = []
      if @exit_pending
        @exit_pending = false
        @mode_announced = false
        due << {role: "system", content: MODE_EXIT}
      end
      if @mode_on
        if !@mode_announced
          @mode_announced = true
          @turns_since_reminder = 0
          due << {role: "system", content: MODE_ENTER}
        elsif @turns_since_reminder >= TURNS_BETWEEN_REFRESHERS
          @turns_since_reminder = 0
          due << {role: "system", content: MODE_REFRESH}
        end
      end
      due
    end
  end
  ```
</CodeGroup>

## Ausführen

<Warning>
  Das Bash-Tool in diesem Beispiel führt vom Modell geschriebene Befehle direkt auf deinem Rechner ohne Sandbox aus, und der Fan-out lässt mehrere dieser Agenten parallel laufen. Führe es in einem Verzeichnis und einer Umgebung aus, die du bedenkenlos offenlegen kannst, und füge Sandboxing hinzu, bevor du es für irgendetwas über lokale Experimente hinaus anpasst.
</Warning>

<CodeGroup exclude="shell:cURL, shell:CLI">
  ```python Python
  if __name__ == "__main__":
      task = (
          sys.argv[1]
          if len(sys.argv) > 1
          else "Explore the current directory, then give a thorough review: what it does, "
          "code-quality issues, and concrete improvements."
      )
      agent = ModeAgent(MODEL)
      print(agent.turn(task))
      agent.set_mode(False)
      print(agent.turn("Briefly summarize what you found above, no fan-out needed."))
  ```

  ```typescript TypeScript
  const task =
    process.argv[2] ??
    "Explore the current directory, then give a thorough review: what it does, " +
      "code-quality issues, and concrete improvements.";
  const agent = new ModeAgent(MODEL);
  console.log(await agent.turn(task));
  agent.setMode(false);
  console.log(await agent.turn("Briefly summarize what you found above, no fan-out needed."));
  ```

  ```csharp C#
  var task = args.Length > 0
      ? args[0]
      : "Explore the current directory, then give a thorough review: what it does, "
          + "code-quality issues, and concrete improvements.";
  Console.WriteLine(await Turn(task));
  SetMode(false);
  Console.WriteLine(await Turn("Briefly summarize what you found above, no fan-out needed."));
  ```

  ```go Go
  func main() {
  	if err := run(context.Background()); err != nil {
  		log.Fatal(err)
  	}
  }

  func run(ctx context.Context) error {
  	if docTestMode {
  		defer os.RemoveAll(workDir)
  	}
  	task := "Explore the current directory, then give a thorough review: what it does, " +
  		"code-quality issues, and concrete improvements."
  	if len(os.Args) > 1 {
  		task = os.Args[1]
  	}
  	agent := newModeAgent(modelID)
  	answer, err := agent.turn(ctx, task)
  	if err != nil {
  		return err
  	}
  	fmt.Println(answer)

  	agent.setMode(false)
  	summary, err := agent.turn(ctx, "Briefly summarize what you found above, no fan-out needed.")
  	if err != nil {
  		return err
  	}
  	fmt.Println(summary)
  	return nil
  }
  ```

  ```java Java
  void main(String[] args) throws InterruptedException {
      String task = args.length > 0 ? args[0] : """
              Explore the current directory, then give a thorough review: what it does, \
              code-quality issues, and concrete improvements.""";
      ModeAgent agent = new ModeAgent(MODEL);
      IO.println(agent.turn(task));
      agent.setMode(false);
      IO.println(agent.turn("Briefly summarize what you found above, no fan-out needed."));
  }
  ```

  ```php PHP
  $task = $argv[1] ??
      'Explore the current directory, then give a thorough review: what it does, '
      . 'code-quality issues, and concrete improvements.';
  $agent = new ModeAgent($client, MODEL);
  echo $agent->turn($task), PHP_EOL;
  $agent->setMode(false);
  echo $agent->turn('Briefly summarize what you found above, no fan-out needed.'), PHP_EOL;
  ```

  ```ruby Ruby
  task = ARGV[0] ||
    "Explore the current directory, then give a thorough review: what it does, " \
    "code-quality issues, and concrete improvements."
  agent = ModeAgent.new(MODEL)
  puts agent.turn(task)
  agent.set_mode(false)
  puts agent.turn("Briefly summarize what you found above, no fan-out needed.")
  ```
</CodeGroup>

Starte das Beispiel aus dem Verzeichnis, in dem die Agenten arbeiten sollen, zum Beispiel dem Wurzelverzeichnis eines zu überprüfenden Repositorys:

```bash
python orchestration_mode.py "Review this repository for flaky tests and propose fixes."
```

Bei eingeschaltetem Modus kannst du erwarten, dass das Modell mit ein paar Bash-Befehlen erkundet, das Workflow-Tool unaufgefordert aufruft und die Subagenten-Berichte zu einer endgültigen Antwort zusammenführt. Triviale oder rein gesprächsartige Anfragen bleiben solo, wie die Erinnerung anweist.

## Auf dem Weg zu einem Produktions-Harness

Dieses Beispiel ist bewusst klein gehalten. Ein Harness für echte Workloads würde typischerweise Folgendes ergänzen:

* **Orchestrierungsskripte in einer Sandbox:** Lass das Modell ein kurzes Orchestrierungsprogramm (Verzweigungen, Schleifen und Reduce-Schritte) ausgeben und führe es in einem isolierten Interpreter aus, statt nur eine flache Liste von Teilaufgaben-Strings zu akzeptieren.
* **Dauerhaftes Journaling:** Ersetze die lokale JSON-Datei durch einen Speicher, der Prozessneustarts überlebt und bei gleichzeitigen Schreibern über mehrere Maschinen hinweg sicher ist.
* **Budget-Durchsetzung:** Verfolge die Gesamtzahl der gestarteten Subagenten über die gesamte Sitzung hinweg, nicht nur pro Workflow-Aufruf, und verweigere das Überschreiten einer harten Obergrenze, damit ein außer Kontrolle geratener Plan dein Kontingent nicht erschöpfen kann.

Die Muster in diesem Beispiel (die Modus-Erinnerungen, dauerhafte Zustimmung in der Tool-Beschreibung, Journaling und eine Verifikationswelle) lassen sich unverändert übernehmen; nur das Ausführungssubstrat um sie herum wird robuster.

## Verwandte Themen

<CardGroup cols={2}>
  <Card title="System-Nachrichten mitten im Gespräch" icon="message" href="https://platform.claude.com/docs/de/build-with-claude/mid-conversation-system-messages">
    Der Mechanismus, den die Modus-Erinnerungen verwenden, und wie er mit Prompt-Caching zusammenspielt.
  </Card>

  <Card title="Effort" icon="gauge" href="https://platform.claude.com/docs/de/build-with-claude/effort">
    Die Effort-Stufen, die die API akzeptiert, und wie du eine auswählst.
  </Card>

  <Card title="Tool-Nutzung mit Claude" icon="wrench" href="https://platform.claude.com/docs/de/agents-and-tools/tool-use/overview">
    Tools definieren, Tool-Aufrufe verarbeiten und Tool-Ergebnisse.
  </Card>

  <Card title="Bash-Tool" icon="terminal" href="https://platform.claude.com/docs/de/agents-and-tools/tool-use/bash-tool">
    Das von Anthropic definierte Bash-Tool, das dieses Beispiel lokal ausführt.
  </Card>
</CardGroup>