opencode-swarm-plugin 0.18.0 → 0.20.0

package/src/swarm-orchestrate.ts

@@ -0,0 +1,1869 @@
+/**
+ * Swarm Orchestrate Module - Status tracking and completion handling
+ *
+ * Handles swarm execution lifecycle:
+ * - Initialization and tool availability
+ * - Status tracking and progress reporting
+ * - Completion verification and gates
+ * - Error accumulation and 3-strike detection
+ * - Learning from outcomes
+ *
+ * Key responsibilities:
+ * - swarm_init - Check tools and discover skills
+ * - swarm_status - Query epic progress
+ * - swarm_progress - Report agent progress
+ * - swarm_complete - Verification gate and completion
+ * - swarm_record_outcome - Learning signals
+ * - swarm_broadcast - Mid-task context sharing
+ * - Error accumulation tools
+ * - 3-strike detection for architectural problems
+ */
+
+import { tool } from "@opencode-ai/plugin";
+import { z } from "zod";
+import {
+  type AgentProgress,
+  AgentProgressSchema,
+  type Bead,
+  BeadSchema,
+  type Evaluation,
+  EvaluationSchema,
+  type SpawnedAgent,
+  type SwarmStatus,
+  SwarmStatusSchema,
+} from "./schemas";
+import {
+  getSwarmInbox,
+  releaseSwarmFiles,
+  sendSwarmMessage,
+} from "./streams/swarm-mail";
+import {
+  addStrike,
+  clearStrikes,
+  DEFAULT_LEARNING_CONFIG,
+  type DecompositionStrategy as LearningDecompositionStrategy,
+  ErrorAccumulator,
+  type ErrorType,
+  type FeedbackEvent,
+  formatMemoryStoreOn3Strike,
+  formatMemoryStoreOnSuccess,
+  getArchitecturePrompt,
+  getStrikes,
+  InMemoryStrikeStorage,
+  isStrikedOut,
+  type OutcomeSignals,
+  OutcomeSignalsSchema,
+  outcomeToFeedback,
+  type ScoredOutcome,
+  scoreImplicitFeedback,
+  type StrikeStorage,
+} from "./learning";
+import {
+  checkAllTools,
+  formatToolAvailability,
+  isToolAvailable,
+  warnMissingTool,
+} from "./tool-availability";
+import { listSkills } from "./skills";
+
+// ============================================================================
+// Helper Functions
+// ============================================================================
+
+/**
+ * Query beads for subtasks of an epic
+ */
+async function queryEpicSubtasks(epicId: string): Promise<Bead[]> {
+  // Check if beads is available
+  const beadsAvailable = await isToolAvailable("beads");
+  if (!beadsAvailable) {
+    warnMissingTool("beads");
+    return []; // Return empty - swarm can still function without status tracking
+  }
+
+  const result = await Bun.$`bd list --parent ${epicId} --json`
+    .quiet()
+    .nothrow();
+
+  if (result.exitCode !== 0) {
+    // Don't throw - just return empty and log error prominently
+    console.error(
+      `[swarm] ERROR: Failed to query subtasks for epic ${epicId}:`,
+      result.stderr.toString(),
+    );
+    return [];
+  }
+
+  try {
+    const parsed = JSON.parse(result.stdout.toString());
+    return z.array(BeadSchema).parse(parsed);
+  } catch (error) {
+    if (error instanceof z.ZodError) {
+      console.error(
+        `[swarm] ERROR: Invalid bead data for epic ${epicId}:`,
+        error.message,
+      );
+      return [];
+    }
+    console.error(
+      `[swarm] ERROR: Failed to parse beads for epic ${epicId}:`,
+      error,
+    );
+    throw error;
+  }
+}
+
+/**
+ * Query Agent Mail for swarm thread messages
+ */
+async function querySwarmMessages(
+  projectKey: string,
+  threadId: string,
+): Promise<number> {
+  // Check if agent-mail is available
+  const agentMailAvailable = await isToolAvailable("agent-mail");
+  if (!agentMailAvailable) {
+    // Don't warn here - it's checked elsewhere
+    return 0;
+  }
+
+  try {
+    // Use embedded swarm-mail inbox to count messages in thread
+    const inbox = await getSwarmInbox({
+      projectPath: projectKey,
+      agentName: "coordinator", // Dummy agent name for thread query
+      limit: 5,
+      includeBodies: false,
+    });
+
+    // Count messages that match the thread ID
+    const threadMessages = inbox.messages.filter(
+      (m) => m.thread_id === threadId,
+    );
+    return threadMessages.length;
+  } catch (error) {
+    // Thread might not exist yet, or query failed
+    console.warn(
+      `[swarm] Failed to query swarm messages for thread ${threadId}:`,
+      error,
+    );
+    return 0;
+  }
+}
+
+/**
+ * Format a progress message for Agent Mail
+ */
+function formatProgressMessage(progress: AgentProgress): string {
+  const lines = [
+    `**Status**: ${progress.status}`,
+    progress.progress_percent !== undefined
+      ? `**Progress**: ${progress.progress_percent}%`
+      : null,
+    progress.message ? `**Message**: ${progress.message}` : null,
+    progress.files_touched && progress.files_touched.length > 0
+      ? `**Files touched**:\n${progress.files_touched.map((f) => `- \`${f}\``).join("\n")}`
+      : null,
+    progress.blockers && progress.blockers.length > 0
+      ? `**Blockers**:\n${progress.blockers.map((b) => `- ${b}`).join("\n")}`
+      : null,
+  ];
+
+  return lines.filter(Boolean).join("\n\n");
+}
+
+// ============================================================================
+// Verification Gate
+// ============================================================================
+
+/**
+ * Verification Gate result - tracks each verification step
+ *
+ * Based on the Gate Function from superpowers:
+ * 1. IDENTIFY: What command proves this claim?
+ * 2. RUN: Execute the FULL command (fresh, complete)
+ * 3. READ: Full output, check exit code, count failures
+ * 4. VERIFY: Does output confirm the claim?
+ * 5. ONLY THEN: Make the claim
+ */
+interface VerificationStep {
+  name: string;
+  command: string;
+  passed: boolean;
+  exitCode: number;
+  output?: string;
+  error?: string;
+  skipped?: boolean;
+  skipReason?: string;
+}
+
+interface VerificationGateResult {
+  passed: boolean;
+  steps: VerificationStep[];
+  summary: string;
+  blockers: string[];
+}
+
+/**
+ * UBS scan result schema
+ */
+interface UbsScanResult {
+  exitCode: number;
+  bugs: Array<{
+    file: string;
+    line: number;
+    severity: string;
+    message: string;
+    category: string;
+  }>;
+  summary: {
+    total: number;
+    critical: number;
+    high: number;
+    medium: number;
+    low: number;
+  };
+}
+
+/**
+ * Run UBS scan on files before completion
+ *
+ * @param files - Files to scan
+ * @returns Scan result or null if UBS not available
+ */
+async function runUbsScan(files: string[]): Promise<UbsScanResult | null> {
+  if (files.length === 0) {
+    return null;
+  }
+
+  // Check if UBS is available first
+  const ubsAvailable = await isToolAvailable("ubs");
+  if (!ubsAvailable) {
+    warnMissingTool("ubs");
+    return null;
+  }
+
+  try {
+    // Run UBS scan with JSON output
+    const result = await Bun.$`ubs scan ${files.join(" ")} --json`
+      .quiet()
+      .nothrow();
+
+    const output = result.stdout.toString();
+    if (!output.trim()) {
+      return {
+        exitCode: result.exitCode,
+        bugs: [],
+        summary: { total: 0, critical: 0, high: 0, medium: 0, low: 0 },
+      };
+    }
+
+    try {
+      const parsed = JSON.parse(output);
+
+      // Basic validation of structure
+      if (typeof parsed !== "object" || parsed === null) {
+        throw new Error("UBS output is not an object");
+      }
+      if (!Array.isArray(parsed.bugs)) {
+        console.warn("[swarm] UBS output missing bugs array, using empty");
+      }
+      if (typeof parsed.summary !== "object" || parsed.summary === null) {
+        console.warn("[swarm] UBS output missing summary object, using empty");
+      }
+
+      return {
+        exitCode: result.exitCode,
+        bugs: Array.isArray(parsed.bugs) ? parsed.bugs : [],
+        summary: parsed.summary || {
+          total: 0,
+          critical: 0,
+          high: 0,
+          medium: 0,
+          low: 0,
+        },
+      };
+    } catch (error) {
+      // UBS output wasn't JSON - this is an error condition
+      console.error(
+        `[swarm] CRITICAL: UBS scan failed to parse JSON output because output is malformed:`,
+        error,
+      );
+      console.error(
+        `[swarm] Raw output: ${output}. Try: Run 'ubs doctor' to check installation, verify UBS version with 'ubs --version' (need v1.0.0+), or check if UBS supports --json flag.`,
+      );
+      return {
+        exitCode: result.exitCode,
+        bugs: [],
+        summary: { total: 0, critical: 0, high: 0, medium: 0, low: 0 },
+      };
+    }
+  } catch {
+    return null;
+  }
+}
+
+/**
+ * Run typecheck verification
+ *
+ * Attempts to run TypeScript type checking on the project.
+ * Falls back gracefully if tsc is not available.
+ */
+async function runTypecheckVerification(): Promise<VerificationStep> {
+  const step: VerificationStep = {
+    name: "typecheck",
+    command: "tsc --noEmit",
+    passed: false,
+    exitCode: -1,
+  };
+
+  try {
+    // Check if tsconfig.json exists in current directory
+    const tsconfigExists = await Bun.file("tsconfig.json").exists();
+    if (!tsconfigExists) {
+      step.skipped = true;
+      step.skipReason = "No tsconfig.json found";
+      step.passed = true; // Don't block if no TypeScript
+      return step;
+    }
+
+    const result = await Bun.$`tsc --noEmit`.quiet().nothrow();
+    step.exitCode = result.exitCode;
+    step.passed = result.exitCode === 0;
+
+    if (!step.passed) {
+      step.error = result.stderr.toString().slice(0, 1000); // Truncate for context
+      step.output = result.stdout.toString().slice(0, 1000);
+    }
+  } catch (error) {
+    step.skipped = true;
+    step.skipReason = `tsc not available: ${error instanceof Error ? error.message : String(error)}`;
+    step.passed = true; // Don't block if tsc unavailable
+  }
+
+  return step;
+}
+
+/**
+ * Run test verification for specific files
+ *
+ * Attempts to find and run tests related to the touched files.
+ * Uses common test patterns (*.test.ts, *.spec.ts, __tests__/).
+ */
+async function runTestVerification(
+  filesTouched: string[],
+): Promise<VerificationStep> {
+  const step: VerificationStep = {
+    name: "tests",
+    command: "bun test <related-files>",
+    passed: false,
+    exitCode: -1,
+  };
+
+  if (filesTouched.length === 0) {
+    step.skipped = true;
+    step.skipReason = "No files touched";
+    step.passed = true;
+    return step;
+  }
+
+  // Find test files related to touched files
+  const testPatterns: string[] = [];
+  for (const file of filesTouched) {
+    // Skip if already a test file
+    if (file.includes(".test.") || file.includes(".spec.")) {
+      testPatterns.push(file);
+      continue;
+    }
+
+    // Look for corresponding test file
+    const baseName = file.replace(/\.(ts|tsx|js|jsx)$/, "");
+    testPatterns.push(`${baseName}.test.ts`);
+    testPatterns.push(`${baseName}.test.tsx`);
+    testPatterns.push(`${baseName}.spec.ts`);
+  }
+
+  // Check if any test files exist
+  const existingTests: string[] = [];
+  for (const pattern of testPatterns) {
+    try {
+      const exists = await Bun.file(pattern).exists();
+      if (exists) {
+        existingTests.push(pattern);
+      }
+    } catch {
+      // File doesn't exist, skip
+    }
+  }
+
+  if (existingTests.length === 0) {
+    step.skipped = true;
+    step.skipReason = "No related test files found";
+    step.passed = true;
+    return step;
+  }
+
+  try {
+    step.command = `bun test ${existingTests.join(" ")}`;
+    const result = await Bun.$`bun test ${existingTests}`.quiet().nothrow();
+    step.exitCode = result.exitCode;
+    step.passed = result.exitCode === 0;
+
+    if (!step.passed) {
+      step.error = result.stderr.toString().slice(0, 1000);
+      step.output = result.stdout.toString().slice(0, 1000);
+    }
+  } catch (error) {
+    step.skipped = true;
+    step.skipReason = `Test runner failed: ${error instanceof Error ? error.message : String(error)}`;
+    step.passed = true; // Don't block if test runner unavailable
+  }
+
+  return step;
+}
+
+/**
+ * Run the full Verification Gate
+ *
+ * Implements the Gate Function (IDENTIFY → RUN → READ → VERIFY → CLAIM):
+ * 1. UBS scan (already exists)
+ * 2. Typecheck
+ * 3. Tests for touched files
+ *
+ * All steps must pass (or be skipped with valid reason) to proceed.
+ */
+async function runVerificationGate(
+  filesTouched: string[],
+  skipUbs: boolean = false,
+): Promise<VerificationGateResult> {
+  const steps: VerificationStep[] = [];
+  const blockers: string[] = [];
+
+  // Step 1: UBS scan
+  if (!skipUbs && filesTouched.length > 0) {
+    const ubsResult = await runUbsScan(filesTouched);
+    if (ubsResult) {
+      const ubsStep: VerificationStep = {
+        name: "ubs_scan",
+        command: `ubs scan ${filesTouched.join(" ")}`,
+        passed: ubsResult.summary.critical === 0,
+        exitCode: ubsResult.exitCode,
+      };
+
+      if (!ubsStep.passed) {
+        ubsStep.error = `Found ${ubsResult.summary.critical} critical bugs`;
+        blockers.push(
+          `UBS found ${ubsResult.summary.critical} critical bug(s). Try: Run 'ubs scan ${filesTouched.join(" ")}' to see details, fix critical bugs in reported files, or use skip_ubs_scan=true to bypass (not recommended).`,
+        );
+      }
+
+      steps.push(ubsStep);
+    } else {
+      steps.push({
+        name: "ubs_scan",
+        command: "ubs scan",
+        passed: true,
+        exitCode: 0,
+        skipped: true,
+        skipReason: "UBS not available",
+      });
+    }
+  }
+
+  // Step 2: Typecheck
+  const typecheckStep = await runTypecheckVerification();
+  steps.push(typecheckStep);
+  if (!typecheckStep.passed && !typecheckStep.skipped) {
+    blockers.push(
+      `Typecheck failed: ${typecheckStep.error?.slice(0, 100) || "type errors found"}. Try: Run 'tsc --noEmit' to see full errors, check tsconfig.json configuration, or fix reported type errors in modified files.`,
+    );
+  }
+
+  // Step 3: Tests
+  const testStep = await runTestVerification(filesTouched);
+  steps.push(testStep);
+  if (!testStep.passed && !testStep.skipped) {
+    blockers.push(
+      `Tests failed: ${testStep.error?.slice(0, 100) || "test failures"}. Try: Run 'bun test ${testStep.command.split(" ").slice(2).join(" ")}' to see full output, check test assertions, or fix failing tests in modified files.`,
+    );
+  }
+
+  // Build summary
+  const passedCount = steps.filter((s) => s.passed).length;
+  const skippedCount = steps.filter((s) => s.skipped).length;
+  const failedCount = steps.filter((s) => !s.passed && !s.skipped).length;
+
+  const summary =
+    failedCount === 0
+      ? `Verification passed: ${passedCount} checks passed, ${skippedCount} skipped`
+      : `Verification FAILED: ${failedCount} checks failed, ${passedCount} passed, ${skippedCount} skipped`;
+
+  return {
+    passed: failedCount === 0,
+    steps,
+    summary,
+    blockers,
+  };
+}
+
+/**
+ * Classify failure based on error message heuristics
+ *
+ * Simple pattern matching to categorize why a task failed.
+ * Used when failure_mode is not explicitly provided.
+ *
+ * @param error - Error object or message
+ * @returns FailureMode classification
+ */
+function classifyFailure(error: Error | string): string {
+  const msg = (typeof error === "string" ? error : error.message).toLowerCase();
+
+  if (msg.includes("timeout")) return "timeout";
+  if (msg.includes("conflict") || msg.includes("reservation"))
+    return "conflict";
+  if (msg.includes("validation") || msg.includes("schema")) return "validation";
+  if (msg.includes("context") || msg.includes("token"))
+    return "context_overflow";
+  if (msg.includes("blocked") || msg.includes("dependency"))
+    return "dependency_blocked";
+  if (msg.includes("cancel")) return "user_cancelled";
+
+  // Check for tool failure patterns
+  if (
+    msg.includes("tool") ||
+    msg.includes("command") ||
+    msg.includes("failed to execute")
+  ) {
+    return "tool_failure";
+  }
+
+  return "unknown";
+}
+
+// ============================================================================
+// Global Storage
+// ============================================================================
+
+/**
+ * Global error accumulator for tracking errors across subtasks
+ *
+ * This is a session-level singleton that accumulates errors during
+ * swarm execution for feeding into retry prompts.
+ */
+const globalErrorAccumulator = new ErrorAccumulator();
+
+/**
+ * Global strike storage for tracking consecutive fix failures
+ */
+const globalStrikeStorage: StrikeStorage = new InMemoryStrikeStorage();
+
+// ============================================================================
+// Tool Definitions
+// ============================================================================
+
+/**
+ * Initialize swarm and check tool availability
+ *
+ * Call this at the start of a swarm session to see what tools are available,
+ * what skills exist in the project, and what features will be degraded.
+ *
+ * Skills are automatically discovered from:
+ * - .opencode/skills/
+ * - .claude/skills/
+ * - skills/
+ */
+export const swarm_init = tool({
+  description:
+    "Initialize swarm session: discovers available skills, checks tool availability. ALWAYS call at swarm start.",
+  args: {
+    project_path: tool.schema
+      .string()
+      .optional()
+      .describe("Project path (for Agent Mail init)"),
+  },
+  async execute(args) {
+    // Check all tools
+    const availability = await checkAllTools();
+
+    // Build status report
+    const report = formatToolAvailability(availability);
+
+    // Check critical tools
+    const beadsAvailable = availability.get("beads")?.status.available ?? false;
+    const agentMailAvailable =
+      availability.get("agent-mail")?.status.available ?? false;
+
+    // Build warnings
+    const warnings: string[] = [];
+    const degradedFeatures: string[] = [];
+
+    if (!beadsAvailable) {
+      warnings.push(
+        "⚠️  beads (bd) not available - issue tracking disabled, swarm coordination will be limited",
+      );
+      degradedFeatures.push("issue tracking", "progress persistence");
+    }
+
+    if (!agentMailAvailable) {
+      warnings.push(
+        "⚠️  agent-mail not available - multi-agent communication disabled",
+      );
+      degradedFeatures.push("agent communication", "file reservations");
+    }
+
+    if (!availability.get("cass")?.status.available) {
+      degradedFeatures.push("historical context from past sessions");
+    }
+
+    if (!availability.get("ubs")?.status.available) {
+      degradedFeatures.push("pre-completion bug scanning");
+    }
+
+    if (!availability.get("semantic-memory")?.status.available) {
+      degradedFeatures.push("persistent learning (using in-memory fallback)");
+    }
+
+    // Discover available skills
+    const availableSkills = await listSkills();
+    const skillsInfo = {
+      count: availableSkills.length,
+      available: availableSkills.length > 0,
+      skills: availableSkills.map((s) => ({
+        name: s.name,
+        description: s.description,
+        hasScripts: s.hasScripts,
+      })),
+    };
+
+    // Add skills guidance if available
+    let skillsGuidance: string | undefined;
+    if (availableSkills.length > 0) {
+      skillsGuidance = `Found ${availableSkills.length} skill(s). Use skills_list to see details, skills_use to activate.`;
+    } else {
+      skillsGuidance =
+        "No skills found. Add skills to .opencode/skills/ or .claude/skills/ for specialized guidance.";
+    }
+
+    return JSON.stringify(
+      {
+        ready: true,
+        tool_availability: Object.fromEntries(
+          Array.from(availability.entries()).map(([k, v]) => [
+            k,
+            {
+              available: v.status.available,
+              fallback: v.status.available ? null : v.fallbackBehavior,
+            },
+          ]),
+        ),
+        skills: skillsInfo,
+        warnings: warnings.length > 0 ? warnings : undefined,
+        degraded_features:
+          degradedFeatures.length > 0 ? degradedFeatures : undefined,
+        recommendations: {
+          skills: skillsGuidance,
+          beads: beadsAvailable
+            ? "✓ Use beads for all task tracking"
+            : "Install beads: npm i -g @joelhooks/beads",
+          agent_mail: agentMailAvailable
+            ? "✓ Use Agent Mail for coordination"
+            : "Start Agent Mail: agent-mail serve",
+        },
+        report,
+      },
+      null,
+      2,
+    );
+  },
+});
+
+/**
+ * Get status of a swarm by epic ID
+ *
+ * Requires project_key to query Agent Mail for message counts.
+ */
+export const swarm_status = tool({
+  description: "Get status of a swarm by epic ID",
+  args: {
+    epic_id: tool.schema.string().describe("Epic bead ID (e.g., bd-abc123)"),
+    project_key: tool.schema
+      .string()
+      .describe("Project path (for Agent Mail queries)"),
+  },
+  async execute(args) {
+    // Query subtasks from beads
+    const subtasks = await queryEpicSubtasks(args.epic_id);
+
+    // Count statuses
+    const statusCounts = {
+      running: 0,
+      completed: 0,
+      failed: 0,
+      blocked: 0,
+    };
+
+    const agents: SpawnedAgent[] = [];
+
+    for (const bead of subtasks) {
+      // Map bead status to agent status
+      let agentStatus: SpawnedAgent["status"] = "pending";
+      switch (bead.status) {
+        case "in_progress":
+          agentStatus = "running";
+          statusCounts.running++;
+          break;
+        case "closed":
+          agentStatus = "completed";
+          statusCounts.completed++;
+          break;
+        case "blocked":
+          agentStatus = "pending"; // Blocked treated as pending for swarm
+          statusCounts.blocked++;
+          break;
+        default:
+          // open = pending
+          break;
+      }
+
+      agents.push({
+        bead_id: bead.id,
+        agent_name: "", // We don't track this in beads
+        status: agentStatus,
+        files: [], // Would need to parse from description
+      });
+    }
+
+    // Query Agent Mail for message activity
+    const messageCount = await querySwarmMessages(
+      args.project_key,
+      args.epic_id,
+    );
+
+    const status: SwarmStatus = {
+      epic_id: args.epic_id,
+      total_agents: subtasks.length,
+      running: statusCounts.running,
+      completed: statusCounts.completed,
+      failed: statusCounts.failed,
+      blocked: statusCounts.blocked,
+      agents,
+      last_update: new Date().toISOString(),
+    };
+
+    // Validate and return
+    const validated = SwarmStatusSchema.parse(status);
+
+    return JSON.stringify(
+      {
+        ...validated,
+        message_count: messageCount,
+        progress_percent:
+          subtasks.length > 0
+            ? Math.round((statusCounts.completed / subtasks.length) * 100)
+            : 0,
+      },
+      null,
+      2,
+    );
+  },
+});
+
+/**
+ * Report progress on a subtask
+ *
+ * Takes explicit agent identity since tools don't have persistent state.
+ */
+export const swarm_progress = tool({
+  description: "Report progress on a subtask to coordinator",
+  args: {
+    project_key: tool.schema.string().describe("Project path"),
+    agent_name: tool.schema.string().describe("Your Agent Mail name"),
+    bead_id: tool.schema.string().describe("Subtask bead ID"),
+    status: tool.schema
+      .enum(["in_progress", "blocked", "completed", "failed"])
+      .describe("Current status"),
+    message: tool.schema
+      .string()
+      .optional()
+      .describe("Progress message or blockers"),
+    progress_percent: tool.schema
+      .number()
+      .min(0)
+      .max(100)
+      .optional()
+      .describe("Completion percentage"),
+    files_touched: tool.schema
+      .array(tool.schema.string())
+      .optional()
+      .describe("Files modified so far"),
+  },
+  async execute(args) {
+    // Build progress report
+    const progress: AgentProgress = {
+      bead_id: args.bead_id,
+      agent_name: args.agent_name,
+      status: args.status,
+      progress_percent: args.progress_percent,
+      message: args.message,
+      files_touched: args.files_touched,
+      timestamp: new Date().toISOString(),
+    };
+
+    // Validate
+    const validated = AgentProgressSchema.parse(progress);
+
+    // Update bead status if needed
+    if (args.status === "blocked" || args.status === "in_progress") {
+      const beadStatus = args.status === "blocked" ? "blocked" : "in_progress";
+      await Bun.$`bd update ${args.bead_id} --status ${beadStatus} --json`
+        .quiet()
+        .nothrow();
+    }
+
+    // Extract epic ID from bead ID (e.g., bd-abc123.1 -> bd-abc123)
+    const epicId = args.bead_id.includes(".")
+      ? args.bead_id.split(".")[0]
+      : args.bead_id;
+
+    // Send progress message to thread using embedded swarm-mail
+    await sendSwarmMessage({
+      projectPath: args.project_key,
+      fromAgent: args.agent_name,
+      toAgents: [], // Coordinator will pick it up from thread
+      subject: `Progress: ${args.bead_id} - ${args.status}`,
+      body: formatProgressMessage(validated),
+      threadId: epicId,
+      importance: args.status === "blocked" ? "high" : "normal",
+    });
+
+    return `Progress reported: ${args.status}${args.progress_percent !== undefined ? ` (${args.progress_percent}%)` : ""}`;
+  },
+});
+
+/**
+ * Broadcast context updates to all agents in the epic
+ *
+ * Enables mid-task coordination by sharing discoveries, warnings, or blockers
+ * with all agents working on the same epic. Agents can broadcast without
+ * waiting for task completion.
+ *
+ * Based on "Patterns for Building AI Agents" p.31: "Ensure subagents can share context along the way"
+ */
+export const swarm_broadcast = tool({
+  description:
+    "Broadcast context update to all agents working on the same epic",
+  args: {
+    project_path: tool.schema
+      .string()
+      .describe("Absolute path to project root"),
+    agent_name: tool.schema
+      .string()
+      .describe("Name of the agent broadcasting the message"),
+    epic_id: tool.schema.string().describe("Epic ID (e.g., bd-abc123)"),
+    message: tool.schema
+      .string()
+      .describe("Context update to share (what changed, what was learned)"),
+    importance: tool.schema
+      .enum(["info", "warning", "blocker"])
+      .default("info")
+      .describe("Priority level (default: info)"),
+    files_affected: tool.schema
+      .array(tool.schema.string())
+      .optional()
+      .describe("Files this context relates to"),
+  },
+  async execute(args) {
+    // Extract bead_id from context if available (for traceability)
+    const beadId = "unknown"; // Context not currently available in tool execution
+
+    // Format the broadcast message
+    const body = [
+      `## Context Update`,
+      "",
+      `**From**: ${args.agent_name} (${beadId})`,
+      `**Priority**: ${args.importance.toUpperCase()}`,
+      "",
+      args.message,
+      "",
+      args.files_affected && args.files_affected.length > 0
+        ? `**Files affected**:\n${args.files_affected.map((f) => `- \`${f}\``).join("\n")}`
+        : "",
+    ]
+      .filter(Boolean)
+      .join("\n");
+
+    // Map importance to Agent Mail importance
+    const mailImportance =
+      args.importance === "blocker"
+        ? "urgent"
+        : args.importance === "warning"
+          ? "high"
+          : "normal";
+
+    // Send as broadcast to thread using embedded swarm-mail
+    await sendSwarmMessage({
+      projectPath: args.project_path,
+      fromAgent: args.agent_name,
+      toAgents: [], // Broadcast to thread
+      subject: `[${args.importance.toUpperCase()}] Context update from ${args.agent_name}`,
+      body,
+      threadId: args.epic_id,
+      importance: mailImportance,
+      ackRequired: args.importance === "blocker",
+    });
+
+    return JSON.stringify(
+      {
+        broadcast: true,
+        epic_id: args.epic_id,
+        from: args.agent_name,
+        bead_id: beadId,
+        importance: args.importance,
+        recipients: "all agents in epic",
+        ack_required: args.importance === "blocker",
+      },
+      null,
+      2,
+    );
+  },
+});
+
+/**
+ * Mark a subtask as complete
+ *
+ * Implements the Verification Gate (from superpowers):
+ * 1. IDENTIFY: What commands prove this claim?
+ * 2. RUN: Execute verification (UBS, typecheck, tests)
+ * 3. READ: Check exit codes and output
+ * 4. VERIFY: All checks must pass
+ * 5. ONLY THEN: Close the bead
+ *
+ * Closes bead, releases reservations, notifies coordinator.
+ */
+export const swarm_complete = tool({
+  description:
+    "Mark subtask complete with Verification Gate. Runs UBS scan, typecheck, and tests before allowing completion.",
+  args: {
+    project_key: tool.schema.string().describe("Project path"),
+    agent_name: tool.schema.string().describe("Your Agent Mail name"),
+    bead_id: tool.schema.string().describe("Subtask bead ID"),
+    summary: tool.schema.string().describe("Brief summary of work done"),
+    evaluation: tool.schema
+      .string()
+      .optional()
+      .describe("Self-evaluation JSON (Evaluation schema)"),
+    files_touched: tool.schema
+      .array(tool.schema.string())
+      .optional()
+      .describe("Files modified - will be verified (UBS, typecheck, tests)"),
+    skip_ubs_scan: tool.schema
+      .boolean()
+      .optional()
+      .describe("Skip UBS bug scan (default: false)"),
+    skip_verification: tool.schema
+      .boolean()
+      .optional()
+      .describe(
+        "Skip ALL verification (UBS, typecheck, tests). Use sparingly! (default: false)",
+      ),
+  },
+  async execute(args) {
+    // Run Verification Gate unless explicitly skipped
+    let verificationResult: VerificationGateResult | null = null;
+
+    if (!args.skip_verification && args.files_touched?.length) {
+      verificationResult = await runVerificationGate(
+        args.files_touched,
+        args.skip_ubs_scan ?? false,
+      );
+
+      // Block completion if verification failed
+      if (!verificationResult.passed) {
+        return JSON.stringify(
+          {
+            success: false,
+            error: "Verification Gate FAILED - fix issues before completing",
+            verification: {
+              passed: false,
+              summary: verificationResult.summary,
+              blockers: verificationResult.blockers,
+              steps: verificationResult.steps.map((s) => ({
+                name: s.name,
+                passed: s.passed,
+                skipped: s.skipped,
+                skipReason: s.skipReason,
+                error: s.error?.slice(0, 200),
+              })),
+            },
+            hint:
+              verificationResult.blockers.length > 0
+                ? `Fix these issues: ${verificationResult.blockers.map((b, i) => `${i + 1}. ${b}`).join(", ")}. Use skip_verification=true only as last resort.`
+                : "Fix the failing checks and try again. Use skip_verification=true only as last resort.",
+            gate_function:
+              "IDENTIFY → RUN → READ → VERIFY → CLAIM (you are at VERIFY, claim blocked)",
+          },
+          null,
+          2,
+        );
+      }
+    }
+
+    // Legacy UBS-only path for backward compatibility (when no files_touched)
+    let ubsResult: UbsScanResult | null = null;
+    if (
+      !args.skip_verification &&
+      !verificationResult &&
+      args.files_touched?.length &&
+      !args.skip_ubs_scan
+    ) {
+      ubsResult = await runUbsScan(args.files_touched);
+
+      // Block completion if critical bugs found
+      if (ubsResult && ubsResult.summary.critical > 0) {
+        return JSON.stringify(
+          {
+            success: false,
+            error: `UBS found ${ubsResult.summary.critical} critical bug(s) that must be fixed before completing`,
+            ubs_scan: {
+              critical_count: ubsResult.summary.critical,
+              bugs: ubsResult.bugs.filter((b) => b.severity === "critical"),
+            },
+            hint: `Fix these critical bugs: ${ubsResult.bugs
+              .filter((b) => b.severity === "critical")
+              .map((b) => `${b.file}:${b.line} - ${b.message}`)
+              .slice(0, 3)
+              .join(
+                "; ",
+              )}. Try: Run 'ubs scan ${args.files_touched?.join(" ") || "."} --json' for full report, fix reported issues, or use skip_ubs_scan=true to bypass (not recommended).`,
+          },
+          null,
+          2,
+        );
+      }
+    }
+
+    // Parse and validate evaluation if provided
+    let parsedEvaluation: Evaluation | undefined;
+    if (args.evaluation) {
+      try {
+        parsedEvaluation = EvaluationSchema.parse(JSON.parse(args.evaluation));
+      } catch (error) {
+        return JSON.stringify(
+          {
+            success: false,
+            error: "Invalid evaluation format",
+            details: error instanceof z.ZodError ? error.issues : String(error),
+          },
+          null,
+          2,
+        );
+      }
+
+      // If evaluation failed, don't complete
+      if (!parsedEvaluation.passed) {
+        return JSON.stringify(
+          {
+            success: false,
+            error: "Self-evaluation failed",
+            retry_suggestion: parsedEvaluation.retry_suggestion,
+            feedback: parsedEvaluation.overall_feedback,
+          },
+          null,
+          2,
+        );
+      }
+    }
+
+    // Close the bead
+    const closeResult =
+      await Bun.$`bd close ${args.bead_id} --reason ${args.summary} --json`
+        .quiet()
+        .nothrow();
+
+    if (closeResult.exitCode !== 0) {
+      throw new Error(
+        `Failed to close bead because bd close command failed: ${closeResult.stderr.toString()}. Try: Verify bead exists and is not already closed with 'bd show ${args.bead_id}', check if bead ID is correct with 'beads_query()', or use beads_close tool directly.`,
+      );
+    }
+
+    // Release file reservations for this agent using embedded swarm-mail
+    try {
+      await releaseSwarmFiles({
+        projectPath: args.project_key,
+        agentName: args.agent_name,
+        // Release all reservations for this agent
+      });
+    } catch (error) {
+      // Release might fail (e.g., no reservations existed)
+      // This is non-fatal - log and continue
+      console.warn(
+        `[swarm] Failed to release file reservations for ${args.agent_name}:`,
+        error,
+      );
+    }
+
+    // Extract epic ID
+    const epicId = args.bead_id.includes(".")
+      ? args.bead_id.split(".")[0]
+      : args.bead_id;
+
+    // Send completion message using embedded swarm-mail
+    const completionBody = [
+      `## Subtask Complete: ${args.bead_id}`,
+      "",
+      `**Summary**: ${args.summary}`,
+      "",
+      parsedEvaluation
+        ? `**Self-Evaluation**: ${parsedEvaluation.passed ? "PASSED" : "FAILED"}`
+        : "",
+      parsedEvaluation?.overall_feedback
+        ? `**Feedback**: ${parsedEvaluation.overall_feedback}`
+        : "",
+    ]
+      .filter(Boolean)
+      .join("\n");
+
+    await sendSwarmMessage({
+      projectPath: args.project_key,
+      fromAgent: args.agent_name,
+      toAgents: [], // Thread broadcast
+      subject: `Complete: ${args.bead_id}`,
+      body: completionBody,
+      threadId: epicId,
+      importance: "normal",
+    });
+
+    // Build success response with semantic-memory integration
+    const response = {
+      success: true,
+      bead_id: args.bead_id,
+      closed: true,
+      reservations_released: true,
+      message_sent: true,
+      verification_gate: verificationResult
+        ? {
+            passed: true,
+            summary: verificationResult.summary,
+            steps: verificationResult.steps.map((s) => ({
+              name: s.name,
+              passed: s.passed,
+              skipped: s.skipped,
+              skipReason: s.skipReason,
+            })),
+          }
+        : args.skip_verification
+          ? { skipped: true, reason: "skip_verification=true" }
+          : { skipped: true, reason: "no files_touched provided" },
+      ubs_scan: ubsResult
+        ? {
+            ran: true,
+            bugs_found: ubsResult.summary.total,
+            summary: ubsResult.summary,
+            warnings: ubsResult.bugs.filter((b) => b.severity !== "critical"),
+          }
+        : verificationResult
+          ? { ran: true, included_in_verification_gate: true }
+          : {
+              ran: false,
+              reason: args.skip_ubs_scan
+                ? "skipped"
+                : "no files or ubs unavailable",
+            },
+      learning_prompt: `## Reflection
+
+Did you learn anything reusable during this subtask? Consider:
+
+1. **Patterns**: Any code patterns or approaches that worked well?
+2. **Gotchas**: Edge cases or pitfalls to warn future agents about?
+3. **Best Practices**: Domain-specific guidelines worth documenting?
+4. **Tool Usage**: Effective ways to use tools for this type of task?
+
+If you discovered something valuable, use \`swarm_learn\` or \`skills_create\` to preserve it as a skill for future swarms.
+
+Files touched: ${args.files_touched?.join(", ") || "none recorded"}`,
+      // Add semantic-memory integration on success
+      memory_store: formatMemoryStoreOnSuccess(
+        args.bead_id,
+        args.summary,
+        args.files_touched || [],
+      ),
+    };
+
+    return JSON.stringify(response, null, 2);
+  },
+});
+
+/**
+ * Record outcome signals from a completed subtask
+ *
+ * Tracks implicit feedback (duration, errors, retries) to score
+ * decomposition quality over time. This data feeds into criterion
+ * weight calculations.
+ *
+ * Strategy tracking enables learning about which decomposition strategies
+ * work best for different task types.
+ *
+ * @see src/learning.ts for scoring logic
+ */
+export const swarm_record_outcome = tool({
+  description:
+    "Record subtask outcome for implicit feedback scoring. Tracks duration, errors, retries to learn decomposition quality.",
+  args: {
+    bead_id: tool.schema.string().describe("Subtask bead ID"),
+    duration_ms: tool.schema
+      .number()
+      .int()
+      .min(0)
+      .describe("Duration in milliseconds"),
+    error_count: tool.schema
+      .number()
+      .int()
+      .min(0)
+      .default(0)
+      .describe("Number of errors encountered"),
+    retry_count: tool.schema
+      .number()
+      .int()
+      .min(0)
+      .default(0)
+      .describe("Number of retry attempts"),
+    success: tool.schema.boolean().describe("Whether the subtask succeeded"),
+    files_touched: tool.schema
+      .array(tool.schema.string())
+      .optional()
+      .describe("Files that were modified"),
+    criteria: tool.schema
+      .array(tool.schema.string())
+      .optional()
+      .describe(
+        "Criteria to generate feedback for (default: all default criteria)",
+      ),
+    strategy: tool.schema
+      .enum(["file-based", "feature-based", "risk-based", "research-based"])
+      .optional()
+      .describe("Decomposition strategy used for this task"),
+    failure_mode: tool.schema
+      .enum([
+        "timeout",
+        "conflict",
+        "validation",
+        "tool_failure",
+        "context_overflow",
+        "dependency_blocked",
+        "user_cancelled",
+        "unknown",
+      ])
+      .optional()
+      .describe(
+        "Failure classification (only when success=false). Auto-classified if not provided.",
+      ),
+    failure_details: tool.schema
+      .string()
+      .optional()
+      .describe("Detailed failure context (error message, stack trace, etc.)"),
+  },
+  async execute(args) {
+    // Build outcome signals
+    const signals: OutcomeSignals = {
+      bead_id: args.bead_id,
+      duration_ms: args.duration_ms,
+      error_count: args.error_count ?? 0,
+      retry_count: args.retry_count ?? 0,
+      success: args.success,
+      files_touched: args.files_touched ?? [],
+      timestamp: new Date().toISOString(),
+      strategy: args.strategy as LearningDecompositionStrategy | undefined,
+      failure_mode: args.failure_mode,
+      failure_details: args.failure_details,
+    };
+
+    // If task failed but no failure_mode provided, try to classify from failure_details
+    if (!args.success && !args.failure_mode && args.failure_details) {
+      const classified = classifyFailure(args.failure_details);
+      signals.failure_mode = classified as OutcomeSignals["failure_mode"];
+    }
+
+    // Validate signals
+    const validated = OutcomeSignalsSchema.parse(signals);
+
+    // Score the outcome
+    const scored: ScoredOutcome = scoreImplicitFeedback(
+      validated,
+      DEFAULT_LEARNING_CONFIG,
+    );
+
+    // Get error patterns from accumulator
+    const errorStats = await globalErrorAccumulator.getErrorStats(args.bead_id);
+
+    // Generate feedback events for each criterion
+    const criteriaToScore = args.criteria ?? [
+      "type_safe",
+      "no_bugs",
+      "patterns",
+      "readable",
+    ];
+    const feedbackEvents: FeedbackEvent[] = criteriaToScore.map((criterion) => {
+      const event = outcomeToFeedback(scored, criterion);
+      // Include strategy in feedback context for future analysis
+      if (args.strategy) {
+        event.context =
+          `${event.context || ""} [strategy: ${args.strategy}]`.trim();
+      }
+      // Include error patterns in feedback context
+      if (errorStats.total > 0) {
+        const errorSummary = Object.entries(errorStats.by_type)
+          .map(([type, count]) => `${type}:${count}`)
+          .join(", ");
+        event.context =
+          `${event.context || ""} [errors: ${errorSummary}]`.trim();
+      }
+      return event;
+    });
+
+    return JSON.stringify(
+      {
+        success: true,
+        outcome: {
+          signals: validated,
+          scored: {
+            type: scored.type,
+            decayed_value: scored.decayed_value,
+            reasoning: scored.reasoning,
+          },
+        },
+        feedback_events: feedbackEvents,
+        error_patterns: errorStats,
+        summary: {
+          feedback_type: scored.type,
+          duration_seconds: Math.round(args.duration_ms / 1000),
+          error_count: args.error_count ?? 0,
+          retry_count: args.retry_count ?? 0,
+          success: args.success,
+          strategy: args.strategy,
+          failure_mode: validated.failure_mode,
+          failure_details: validated.failure_details,
+          accumulated_errors: errorStats.total,
+          unresolved_errors: errorStats.unresolved,
+        },
+        note: "Feedback events should be stored for criterion weight calculation. Use learning.ts functions to apply weights.",
+      },
+      null,
+      2,
+    );
+  },
+});
+
+/**
+ * Record an error during subtask execution
+ *
+ * Implements pattern from "Patterns for Building AI Agents" p.40:
+ * "Good agents examine and correct errors when something goes wrong"
+ *
+ * Errors are accumulated and can be fed into retry prompts to help
+ * agents learn from past failures.
+ */
+export const swarm_accumulate_error = tool({
+  description:
+    "Record an error during subtask execution. Errors feed into retry prompts.",
+  args: {
+    bead_id: tool.schema.string().describe("Bead ID where error occurred"),
+    error_type: tool.schema
+      .enum(["validation", "timeout", "conflict", "tool_failure", "unknown"])
+      .describe("Category of error"),
+    message: tool.schema.string().describe("Human-readable error message"),
+    stack_trace: tool.schema
+      .string()
+      .optional()
+      .describe("Stack trace for debugging"),
+    tool_name: tool.schema.string().optional().describe("Tool that failed"),
+    context: tool.schema
+      .string()
+      .optional()
+      .describe("What was happening when error occurred"),
+  },
+  async execute(args) {
+    const entry = await globalErrorAccumulator.recordError(
+      args.bead_id,
+      args.error_type as ErrorType,
+      args.message,
+      {
+        stack_trace: args.stack_trace,
+        tool_name: args.tool_name,
+        context: args.context,
+      },
+    );
+
+    return JSON.stringify(
+      {
+        success: true,
+        error_id: entry.id,
+        bead_id: entry.bead_id,
+        error_type: entry.error_type,
+        message: entry.message,
+        timestamp: entry.timestamp,
+        note: "Error recorded for retry context. Use swarm_get_error_context to retrieve accumulated errors.",
+      },
+      null,
+      2,
+    );
+  },
+});
+
+/**
+ * Get accumulated errors for a bead to feed into retry prompts
+ *
+ * Returns formatted error context that can be injected into retry prompts
+ * to help agents learn from past failures.
+ */
+export const swarm_get_error_context = tool({
+  description:
+    "Get accumulated errors for a bead. Returns formatted context for retry prompts.",
+  args: {
+    bead_id: tool.schema.string().describe("Bead ID to get errors for"),
+    include_resolved: tool.schema
+      .boolean()
+      .optional()
+      .describe("Include resolved errors (default: false)"),
+  },
+  async execute(args) {
+    const errorContext = await globalErrorAccumulator.getErrorContext(
+      args.bead_id,
+      args.include_resolved ?? false,
+    );
+
+    const stats = await globalErrorAccumulator.getErrorStats(args.bead_id);
+
+    return JSON.stringify(
+      {
+        bead_id: args.bead_id,
+        error_context: errorContext,
+        stats: {
+          total_errors: stats.total,
+          unresolved: stats.unresolved,
+          by_type: stats.by_type,
+        },
+        has_errors: errorContext.length > 0,
+        usage:
+          "Inject error_context into retry prompt using {error_context} placeholder",
+      },
+      null,
+      2,
+    );
+  },
+});
+
+/**
+ * Mark an error as resolved
+ *
+ * Call this after an agent successfully addresses an error to update
+ * the accumulator state.
+ */
+export const swarm_resolve_error = tool({
+  description:
+    "Mark an error as resolved after fixing it. Updates error accumulator state.",
+  args: {
+    error_id: tool.schema.string().describe("Error ID to mark as resolved"),
+  },
+  async execute(args) {
+    await globalErrorAccumulator.resolveError(args.error_id);
+
+    return JSON.stringify(
+      {
+        success: true,
+        error_id: args.error_id,
+        resolved: true,
+      },
+      null,
+      2,
+    );
+  },
+});
+
+/**
+ * Check if a bead has struck out (3 consecutive failures)
+ *
+ * The 3-Strike Rule:
+ * IF 3+ fixes have failed:
+ *   STOP → Question the architecture
+ *   DON'T attempt Fix #4
+ *   Discuss with human partner
+ *
+ * This is NOT a failed hypothesis.
+ * This is a WRONG ARCHITECTURE.
+ *
+ * Use this tool to:
+ * - Check strike count before attempting a fix
+ * - Get architecture review prompt if struck out
+ * - Record a strike when a fix fails
+ * - Clear strikes when a fix succeeds
+ */
+export const swarm_check_strikes = tool({
+  description:
+    "Check 3-strike status for a bead. Records failures, detects architectural problems, generates architecture review prompts.",
+  args: {
+    bead_id: tool.schema.string().describe("Bead ID to check"),
+    action: tool.schema
+      .enum(["check", "add_strike", "clear", "get_prompt"])
+      .describe(
+        "Action: check count, add strike, clear strikes, or get prompt",
+      ),
+    attempt: tool.schema
+      .string()
+      .optional()
+      .describe("Description of fix attempt (required for add_strike)"),
+    reason: tool.schema
+      .string()
+      .optional()
+      .describe("Why the fix failed (required for add_strike)"),
+  },
+  async execute(args) {
+    switch (args.action) {
+      case "check": {
+        const count = await getStrikes(args.bead_id, globalStrikeStorage);
+        const strikedOut = await isStrikedOut(
+          args.bead_id,
+          globalStrikeStorage,
+        );
+
+        return JSON.stringify(
+          {
+            bead_id: args.bead_id,
+            strike_count: count,
+            is_striked_out: strikedOut,
+            message: strikedOut
+              ? "⚠️ STRUCK OUT: 3 strikes reached. Use get_prompt action for architecture review."
+              : count === 0
+                ? "No strikes. Clear to proceed."
+                : `${count} strike${count > 1 ? "s" : ""}. ${3 - count} remaining before architecture review required.`,
+            next_action: strikedOut
+              ? "Call with action=get_prompt to get architecture review questions"
+              : "Continue with fix attempt",
+          },
+          null,
+          2,
+        );
+      }
+
+      case "add_strike": {
+        if (!args.attempt || !args.reason) {
+          return JSON.stringify(
+            {
+              error: "add_strike requires 'attempt' and 'reason' parameters",
+            },
+            null,
+            2,
+          );
+        }
+
+        const record = await addStrike(
+          args.bead_id,
+          args.attempt,
+          args.reason,
+          globalStrikeStorage,
+        );
+
+        const strikedOut = record.strike_count >= 3;
+
+        // Build response with memory storage hint on 3-strike
+        const response: Record<string, unknown> = {
+          bead_id: args.bead_id,
+          strike_count: record.strike_count,
+          is_striked_out: strikedOut,
+          failures: record.failures,
+          message: strikedOut
+            ? "⚠️ STRUCK OUT: 3 strikes reached. STOP and question the architecture."
+            : `Strike ${record.strike_count} recorded. ${3 - record.strike_count} remaining.`,
+          warning: strikedOut
+            ? "DO NOT attempt Fix #4. Call with action=get_prompt for architecture review."
+            : undefined,
+        };
+
+        // Add semantic-memory storage hint on 3-strike
+        if (strikedOut) {
+          response.memory_store = formatMemoryStoreOn3Strike(
+            args.bead_id,
+            record.failures,
+          );
+        }
+
+        return JSON.stringify(response, null, 2);
+      }
+
+      case "clear": {
+        await clearStrikes(args.bead_id, globalStrikeStorage);
+
+        return JSON.stringify(
+          {
+            bead_id: args.bead_id,
+            strike_count: 0,
+            is_striked_out: false,
+            message: "Strikes cleared. Fresh start.",
+          },
+          null,
+          2,
+        );
+      }
+
+      case "get_prompt": {
+        const prompt = await getArchitecturePrompt(
+          args.bead_id,
+          globalStrikeStorage,
+        );
+
+        if (!prompt) {
+          return JSON.stringify(
+            {
+              bead_id: args.bead_id,
+              has_prompt: false,
+              message: "No architecture prompt (not struck out yet)",
+            },
+            null,
+            2,
+          );
+        }
+
+        return JSON.stringify(
+          {
+            bead_id: args.bead_id,
+            has_prompt: true,
+            architecture_review_prompt: prompt,
+            message:
+              "Architecture review required. Present this prompt to the human partner.",
+          },
+          null,
+          2,
+        );
+      }
+
+      default:
+        return JSON.stringify(
+          {
+            error: `Unknown action: ${args.action}`,
+          },
+          null,
+          2,
+        );
+    }
+  },
+});
+
+/**
+ * Learn from completed work and optionally create a skill
+ *
+ * This tool helps agents reflect on patterns, best practices, or domain
+ * knowledge discovered during task execution and codify them into reusable
+ * skills for future swarms.
+ *
+ * Implements the "learning swarm" pattern where swarms get smarter over time.
+ */
+export const swarm_learn = tool({
+  description: `Analyze completed work and optionally create a skill from learned patterns.
+
+Use after completing a subtask when you've discovered:
+- Reusable code patterns or approaches
+- Domain-specific best practices
+- Gotchas or edge cases to warn about
+- Effective tool usage patterns
+
+This tool helps you formalize learnings into a skill that future agents can discover and use.`,
+  args: {
+    summary: tool.schema
+      .string()
+      .describe("Brief summary of what was learned (1-2 sentences)"),
+    pattern_type: tool.schema
+      .enum([
+        "code-pattern",
+        "best-practice",
+        "gotcha",
+        "tool-usage",
+        "domain-knowledge",
+        "workflow",
+      ])
+      .describe("Category of the learning"),
+    details: tool.schema
+      .string()
+      .describe("Detailed explanation of the pattern or practice"),
+    example: tool.schema
+      .string()
+      .optional()
+      .describe("Code example or concrete illustration"),
+    when_to_use: tool.schema
+      .string()
+      .describe("When should an agent apply this knowledge?"),
+    files_context: tool.schema
+      .array(tool.schema.string())
+      .optional()
+      .describe("Files that exemplify this pattern"),
+    create_skill: tool.schema
+      .boolean()
+      .optional()
+      .describe(
+        "Create a skill from this learning (default: false, just document)",
+      ),
+    skill_name: tool.schema
+      .string()
+      .regex(/^[a-z0-9-]+$/)
+      .max(64)
+      .optional()
+      .describe("Skill name if creating (required if create_skill=true)"),
+    skill_tags: tool.schema
+      .array(tool.schema.string())
+      .optional()
+      .describe("Tags for the skill if creating"),
+  },
+  async execute(args) {
+    // Format the learning as structured documentation
+    const learning = {
+      summary: args.summary,
+      type: args.pattern_type,
+      details: args.details,
+      example: args.example,
+      when_to_use: args.when_to_use,
+      files_context: args.files_context,
+      recorded_at: new Date().toISOString(),
+    };
+
+    // If creating a skill, generate and create it
+    if (args.create_skill) {
+      if (!args.skill_name) {
+        return JSON.stringify(
+          {
+            success: false,
+            error: "skill_name is required when create_skill=true",
+            learning: learning,
+          },
+          null,
+          2,
+        );
+      }
+
+      // Build skill body from learning
+      const skillBody = `# ${args.summary}
+
+## When to Use
+${args.when_to_use}
+
+## ${args.pattern_type.replace(/-/g, " ").replace(/\b\w/g, (c) => c.toUpperCase())}
+
+${args.details}
+
+${args.example ? `## Example\n\n\`\`\`\n${args.example}\n\`\`\`\n` : ""}
+${args.files_context && args.files_context.length > 0 ? `## Reference Files\n\n${args.files_context.map((f) => `- \`${f}\``).join("\n")}\n` : ""}
+
+---
+*Learned from swarm execution on ${new Date().toISOString().split("T")[0]}*`;
+
+      // Import skills_create functionality
+      const { getSkill, invalidateSkillsCache } = await import("./skills");
+      const { mkdir, writeFile } = await import("node:fs/promises");
+      const { join } = await import("node:path");
+
+      // Check if skill exists
+      const existing = await getSkill(args.skill_name);
+      if (existing) {
+        return JSON.stringify(
+          {
+            success: false,
+            error: `Skill '${args.skill_name}' already exists`,
+            existing_path: existing.path,
+            learning: learning,
+            suggestion:
+              "Use skills_update to add to existing skill, or choose a different name",
+          },
+          null,
+          2,
+        );
+      }
+
+      // Create skill directory and file
+      const skillDir = join(
+        process.cwd(),
+        ".opencode",
+        "skills",
+        args.skill_name,
+      );
+      const skillPath = join(skillDir, "SKILL.md");
+
+      const frontmatter = [
+        "---",
+        `name: ${args.skill_name}`,
+        `description: ${args.when_to_use.slice(0, 200)}${args.when_to_use.length > 200 ? "..." : ""}`,
+        "tags:",
+        `  - ${args.pattern_type}`,
+        `  - learned`,
+        ...(args.skill_tags || []).map((t) => `  - ${t}`),
+        "---",
+      ].join("\n");
+
+      try {
+        await mkdir(skillDir, { recursive: true });
+        await writeFile(skillPath, `${frontmatter}\n\n${skillBody}`, "utf-8");
+        invalidateSkillsCache();
+
+        return JSON.stringify(
+          {
+            success: true,
+            skill_created: true,
+            skill: {
+              name: args.skill_name,
+              path: skillPath,
+              type: args.pattern_type,
+            },
+            learning: learning,
+            message: `Created skill '${args.skill_name}' from learned pattern. Future agents can discover it with skills_list.`,
+          },
+          null,
+          2,
+        );
+      } catch (error) {
+        return JSON.stringify(
+          {
+            success: false,
+            error: `Failed to create skill: ${error instanceof Error ? error.message : String(error)}`,
+            learning: learning,
+          },
+          null,
+          2,
+        );
+      }
+    }
+
+    // Just document the learning without creating a skill
+    return JSON.stringify(
+      {
+        success: true,
+        skill_created: false,
+        learning: learning,
+        message:
+          "Learning documented. Use create_skill=true to persist as a skill for future agents.",
+        suggested_skill_name:
+          args.skill_name ||
+          args.summary
+            .toLowerCase()
+            .replace(/[^a-z0-9\s-]/g, "")
+            .replace(/\s+/g, "-")
+            .slice(0, 64),
+      },
+      null,
+      2,
+    );
+  },
+});
+
+// ============================================================================
+// Export tools
+// ============================================================================
+
+export const orchestrateTools = {
+  swarm_init,
+  swarm_status,
+  swarm_progress,
+  swarm_broadcast,
+  swarm_complete,
+  swarm_record_outcome,
+  swarm_accumulate_error,
+  swarm_get_error_context,
+  swarm_resolve_error,
+  swarm_check_strikes,
+  swarm_learn,
+};