@workermill/agent 0.8.1 → 0.8.3

package/dist/config.d.ts

@@ -16,8 +16,6 @@ export interface AgentConfig {
     bitbucketToken: string;
     gitlabToken: string;
     workerImage: string;
-    teamPlanningEnabled: boolean;
-    analystModel?: string;
 }
 export interface FileConfig {
     apiUrl: string;
@@ -32,8 +30,6 @@ export interface FileConfig {
         gitlab: string;
     };
     workerImage: string;
-    teamPlanningEnabled?: boolean;
-    analystModel?: string;
     setupCompletedAt: string;
 }
 export declare function getConfigDir(): string;

package/dist/config.js

@@ -75,8 +75,6 @@ export function loadConfigFromFile() {
         bitbucketToken: fc.tokens?.bitbucket || "",
         gitlabToken: fc.tokens?.gitlab || "",
         workerImage,
-        teamPlanningEnabled: fc.teamPlanningEnabled ?? true,
-        analystModel: fc.analystModel,
     };
 }
 /**
@@ -121,8 +119,6 @@ export function loadConfig() {
         bitbucketToken: process.env.BITBUCKET_TOKEN || "",
         gitlabToken: process.env.GITLAB_TOKEN || "",
         workerImage: process.env.WORKER_IMAGE || "workermill-worker:local",
-        teamPlanningEnabled: process.env.TEAM_PLANNING_ENABLED !== "false",
-        analystModel: process.env.ANALYST_MODEL,
     };
 }
 /**

package/dist/plan-validator.js

@@ -222,6 +222,7 @@ Review this execution plan against the PRD:
 4. **Unrealistic Scope** - Any step targeting >5 files MUST score below 80 (auto-rejection threshold). Each step should modify at most 5 files. If a step needs more, split it into multiple steps first.
 5. **Missing Operational Steps** - If the PRD requires deployment, provisioning, migrations, or running commands, does the plan include operational steps? Writing code is not the same as deploying it.
 6. **Overlapping File Scope** - If two or more steps share the same targetFiles, this causes parallel merge conflicts. Steps MUST NOT overlap on targetFiles. Deduct 10 points per shared file across steps.
+7. **Serialization Bottleneck** - If more than half the stories depend on a single story that targets >5 files, the plan has a bottleneck. Deduct 15 points — split the foundation or allow more parallel work.
 
 ## Scoring Guide
 

package/dist/planner.d.ts

@@ -28,12 +28,13 @@ export interface PlanningTask {
  *
  * Flow:
  *   1. Fetch planning prompt from cloud API
- *   2. Run Claude CLI to generate plan
- *   3. Parse plan, apply file cap (max 5 files per story)
- *   4. Run critic validation via Claude CLI
- *   5. If critic approves (score >= 80): post validated plan to API
- *   6. If critic rejects: re-run planner with feedback (up to MAX_ITERATIONS)
- *   7. After MAX_ITERATIONS without approval: post best plan if score >= 50 (fallback)
- *   8. If no plan scored >= 50: fail the task
+ *   2. Clone target repo (if available) so planner can explore with tools
+ *   3. Run Claude CLI to generate plan
+ *   4. Parse plan, apply file cap (max 5 files per story)
+ *   5. Run critic validation via Claude CLI
+ *   6. If critic approves (score >= 80): post validated plan to API
+ *   7. If critic rejects: re-run planner with feedback (up to MAX_ITERATIONS)
+ *   8. After MAX_ITERATIONS without approval: post best plan if score >= 50 (fallback)
+ *   9. If no plan scored >= 50: fail the task
  */
 export declare function planTask(task: PlanningTask, config: AgentConfig, credentials?: ClaimCredentials): Promise<boolean>;

package/dist/planner.js

@@ -80,9 +80,10 @@ const logQueue = [];
 let logDrainPromise = null;
 async function drainLogQueue() {
     while (logQueue.length > 0) {
-        const entry = logQueue.shift();
+        // Drain up to 50 entries per batch POST
+        const batch = logQueue.splice(0, 50);
         try {
-            await api.post("/api/control-center/logs", entry, { timeout: 5_000 });
+            await api.post("/api/control-center/logs/batch", { entries: batch }, { timeout: 5_000 });
         }
         catch {
             // Best-effort — drop on failure
@@ -157,7 +158,7 @@ function phaseLabel(phase, elapsed) {
  * Run Claude CLI with stream-json output, posting real-time phase milestones
  * to the cloud dashboard — identical terminal experience to cloud planning.
  */
-function runClaudeCli(claudePath, model, prompt, env, taskId, startTime, disableTools = false) {
+function runClaudeCli(claudePath, model, prompt, env, taskId, startTime, cwd) {
     const taskLabel = chalk.cyan(taskId.slice(0, 8));
     return new Promise((resolve, reject) => {
         const cliArgs = [
@@ -167,12 +168,8 @@ function runClaudeCli(claudePath, model, prompt, env, taskId, startTime, disable
             "--model", model,
             "--permission-mode", "bypassPermissions",
         ];
-        // When analysts already explored the repo, strip tools so the planner
-        // doesn't waste turns re-exploring — it has all context in the prompt.
-        if (disableTools) {
-            cliArgs.push("--allowedTools", "");
-        }
         const proc = spawn(claudePath, cliArgs, {
+            cwd,
             env,
             stdio: ["pipe", "pipe", "pipe"],
         });
@@ -222,7 +219,7 @@ function runClaudeCli(claudePath, model, prompt, env, taskId, startTime, disable
             console.log(`${ts()} ${taskLabel} ${chalk.dim(msg)}`);
         }
         // Flush buffered LLM text to dashboard every 1s (complete lines only)
-        const textFlushInterval = setInterval(() => flushTextBuffer(), 1_000);
+        const textFlushInterval = setInterval(() => flushTextBuffer(), 500);
         // SSE progress updates every 2s — drives PlanningTerminalBar in dashboard
         // (same cadence as local dev's progressInterval in planning-agent-local.ts)
         const sseProgressInterval = setInterval(() => {
@@ -422,15 +419,15 @@ function buildCloneUrl(repo, token, scmProvider) {
     }
 }
 /**
- * Clone the target repo to a temp directory for team planning analysis.
- * Returns the path on success, or null on failure (fallback to single-agent).
+ * Clone the target repo to a temp directory so the planner can explore with tools.
+ * Returns the path on success, or null on failure.
  */
 async function cloneTargetRepo(repo, token, scmProvider, taskId) {
     const taskLabel = chalk.cyan(taskId.slice(0, 8));
     const tmpDir = `/tmp/workermill-planning-${taskId.slice(0, 8)}-${Date.now()}`;
     try {
         const cloneUrl = buildCloneUrl(repo, token, scmProvider);
-        console.log(`${ts()} ${taskLabel} ${chalk.dim("Cloning repo for team planning...")}`);
+        console.log(`${ts()} ${taskLabel} ${chalk.dim("Cloning repo for planner...")}`);
         execSync(`git clone --depth 1 --single-branch "${cloneUrl}" "${tmpDir}"`, {
             stdio: "ignore",
             timeout: 60_000,
@@ -440,7 +437,7 @@ async function cloneTargetRepo(repo, token, scmProvider, taskId) {
     }
     catch (error) {
         const errMsg = error instanceof Error ? error.message : String(error);
-        console.error(`${ts()} ${taskLabel} ${chalk.yellow("⚠")} Clone failed, falling back to single-agent: ${errMsg.substring(0, 100)}`);
+        console.error(`${ts()} ${taskLabel} ${chalk.yellow("⚠")} Clone failed, planner will run without repo access: ${errMsg.substring(0, 100)}`);
         // Cleanup partial clone
         try {
             execSync(`rm -rf "${tmpDir}"`, { stdio: "ignore" });
@@ -451,349 +448,19 @@ async function cloneTargetRepo(repo, token, scmProvider, taskId) {
         return null;
     }
 }
-/**
- * Run an analyst agent via Claude CLI with tool access to the cloned repo.
- * Returns the analyst's report text, or an empty string on failure.
- */
-function runAnalyst(name, claudePath, model, prompt, repoPath, env, timeoutMs = 900_000, taskId) {
-    const label = chalk.blue(`[${name}]`);
-    const modelLabel = chalk.yellow(model);
-    return new Promise((resolve) => {
-        console.log(`${ts()} ${label} Starting analyst using ${modelLabel}...`);
-        if (taskId) {
-            postLog(taskId, `${PREFIX} [${name}] Starting analyst using ${model}...`);
-        }
-        const proc = spawn(claudePath, [
-            "--print",
-            "--verbose",
-            "--output-format", "stream-json",
-            "--model", model,
-            "--permission-mode", "bypassPermissions",
-        ], {
-            cwd: repoPath,
-            env,
-            stdio: ["pipe", "pipe", "pipe"],
-        });
-        // Write prompt via stdin (same as runClaudeCli)
-        proc.stdin.write(prompt);
-        proc.stdin.end();
-        let resultText = "";
-        let fullText = "";
-        let stderrOutput = "";
-        let lineBuffer = "";
-        let toolCalls = 0;
-        let timedOut = false;
-        const startMs = Date.now();
-        proc.stderr.on("data", (chunk) => {
-            const text = chunk.toString();
-            stderrOutput += text;
-            // Show stderr in real-time so we can see what's happening
-            for (const line of text.split("\n").filter((l) => l.trim())) {
-                console.log(`${ts()} ${label} ${chalk.red("stderr:")} ${line.trim()}`);
-            }
-        });
-        proc.stdout.on("data", (data) => {
-            lineBuffer += data.toString();
-            const lines = lineBuffer.split("\n");
-            lineBuffer = lines.pop() || "";
-            for (const line of lines) {
-                const trimmed = line.trim();
-                if (!trimmed)
-                    continue;
-                try {
-                    const event = JSON.parse(trimmed);
-                    // Claude CLI stream-json wraps content in assistant message events
-                    if (event.type === "assistant" && event.message?.content) {
-                        const content = event.message.content;
-                        if (Array.isArray(content)) {
-                            for (const block of content) {
-                                if (block.type === "text" && block.text) {
-                                    fullText += block.text;
-                                    // Log analyst reasoning (first line, truncated)
-                                    const thought = block.text.trim().split("\n")[0].substring(0, 120);
-                                    if (thought) {
-                                        console.log(`${ts()} ${label} ${chalk.dim("💭")} ${chalk.dim(thought)}`);
-                                        if (taskId)
-                                            postLog(taskId, `${PREFIX} [${name}] 💭 ${thought}`);
-                                    }
-                                }
-                                else if (block.type === "tool_use") {
-                                    toolCalls++;
-                                    const toolName = block.name || "unknown";
-                                    // Show tool name + input preview (file path, pattern, etc.)
-                                    const inputStr = block.input ? JSON.stringify(block.input) : "";
-                                    const inputPreview = inputStr.length > 80 ? inputStr.substring(0, 80) + "…" : inputStr;
-                                    console.log(`${ts()} ${label} ${chalk.dim(`Tool: ${toolName}`)}${inputPreview ? chalk.dim(` ${inputPreview}`) : ""}`);
-                                    if (taskId)
-                                        postLog(taskId, `${PREFIX} [${name}] Tool: ${toolName} ${inputPreview}`);
-                                }
-                            }
-                        }
-                        else if (typeof content === "string") {
-                            fullText += content;
-                        }
-                    }
-                    else if (event.type === "content_block_delta" && event.delta?.text) {
-                        // Fallback: raw API streaming format (may appear in some CLI versions)
-                        fullText += event.delta.text;
-                    }
-                    else if (event.type === "content_block_start" && event.content_block?.type === "tool_use") {
-                        toolCalls++;
-                        const toolName = event.content_block?.name || "unknown";
-                        console.log(`${ts()} ${label} ${chalk.dim(`Tool: ${toolName}`)}`);
-                        if (taskId)
-                            postLog(taskId, `${PREFIX} [${name}] Tool: ${toolName}`);
-                    }
-                    else if (event.type === "result" && event.result) {
-                        resultText =
-                            typeof event.result === "string" ? event.result : "";
-                    }
-                }
-                catch {
-                    fullText += trimmed + "\n";
-                }
-            }
-        });
-        const timeout = setTimeout(() => {
-            timedOut = true;
-            proc.kill("SIGTERM");
-            const elapsed = Math.round((Date.now() - startMs) / 1000);
-            console.log(`${ts()} ${label} ${chalk.yellow("⚠ Timed out")} after ${formatElapsed(elapsed)} (${toolCalls} tool calls, ${fullText.length} chars)`);
-            if (taskId)
-                postLog(taskId, `${PREFIX} [${name}] ⚠ Timed out after ${formatElapsed(elapsed)}`);
-            resolve(resultText || fullText || "");
-        }, timeoutMs);
-        proc.on("exit", (code) => {
-            clearTimeout(timeout);
-            const elapsed = Math.round((Date.now() - startMs) / 1000);
-            if (timedOut)
-                return; // already resolved
-            const output = resultText || fullText || "";
-            if (code === 0 && output.length > 0) {
-                console.log(`${ts()} ${label} ${chalk.green("✓ Done")} in ${formatElapsed(elapsed)} (${toolCalls} tool calls, ${output.length} chars)`);
-                if (taskId)
-                    postLog(taskId, `${PREFIX} [${name}] ✓ Done in ${formatElapsed(elapsed)} (${toolCalls} tool calls, ${output.length} chars)`);
-            }
-            else if (code !== 0) {
-                console.log(`${ts()} ${label} ${chalk.red(`✗ Exited ${code}`)} after ${formatElapsed(elapsed)} — ${stderrOutput.substring(0, 150) || "no stderr"}`);
-                if (taskId)
-                    postLog(taskId, `${PREFIX} [${name}] ✗ Exited ${code} after ${formatElapsed(elapsed)}`);
-            }
-            else {
-                console.log(`${ts()} ${label} ${chalk.yellow("⚠ Empty output")} after ${formatElapsed(elapsed)} (${toolCalls} tool calls)`);
-                if (taskId)
-                    postLog(taskId, `${PREFIX} [${name}] ⚠ Empty output after ${formatElapsed(elapsed)}`);
-            }
-            resolve(output);
-        });
-        proc.on("error", (err) => {
-            clearTimeout(timeout);
-            console.log(`${ts()} ${label} ${chalk.red("✗ Spawn failed:")} ${err.message}`);
-            resolve("");
-        });
-    });
-}
-/**
- * Run an analyst agent via Vercel AI SDK with tool access to the cloned repo.
- * Used for non-Anthropic providers (OpenAI, Google, Ollama) that can't use Claude CLI.
- * Returns the analyst's report text, or an empty string on failure.
- */
-async function runAnalystWithSdk(name, provider, model, apiKey, prompt, repoPath, timeoutMs = 900_000, taskId) {
-    const label = chalk.blue(`[${name}]`);
-    const modelLabel = chalk.yellow(`${provider}/${model}`);
-    const startMs = Date.now();
-    console.log(`${ts()} ${label} Starting analyst using ${modelLabel} via AI SDK...`);
-    if (taskId)
-        postLog(taskId, `${PREFIX} [${name}] Starting analyst using ${provider}/${model} via AI SDK...`);
-    try {
-        const result = await generateTextWithTools({
-            provider,
-            model,
-            apiKey,
-            prompt,
-            workingDir: repoPath,
-            maxTokens: 16384,
-            temperature: 0.3,
-            timeoutMs,
-            maxSteps: 20, // Allow thorough exploration
-            enableTools: true,
-        });
-        const elapsed = Math.round((Date.now() - startMs) / 1000);
-        if (result && result.length > 0) {
-            console.log(`${ts()} ${label} ${chalk.green("✓ Done")} in ${elapsed}s (${result.length} chars)`);
-            return result;
-        }
-        console.log(`${ts()} ${label} ${chalk.yellow("⚠ Empty output")} after ${elapsed}s`);
-        return "";
-    }
-    catch (error) {
-        const elapsed = Math.round((Date.now() - startMs) / 1000);
-        const errMsg = error instanceof Error ? error.message : String(error);
-        console.log(`${ts()} ${label} ${chalk.red(`✗ Failed`)} after ${elapsed}s: ${errMsg.substring(0, 150)}`);
-        return "";
-    }
-}
-/** Analyst prompt templates */
-const CODEBASE_ANALYST_PROMPT = `You are a codebase analyst. Your job is to explore this repository using tools and report what you find.
-
-IMPORTANT: You MUST use tools to explore the repository. Do NOT guess or make assumptions.
-
-Step 1: Run Glob with pattern "**/*" to see the top-level directory structure.
-Step 2: Read key files: package.json, tsconfig.json, README.md, .env.example, or equivalents.
-Step 3: Run Glob on src/ or the main source directory to understand the code layout.
-Step 4: Read 2-3 representative source files to understand patterns and frameworks.
-
-After exploring, write a report covering:
-1. Directory structure and organization
-2. Languages, frameworks, and key dependencies (from package.json, requirements.txt, etc.)
-3. Existing test files and testing patterns (search for test/, __tests__, *.test.*, *.spec.*)
-4. CI/CD configuration (search for .github/workflows/, Jenkinsfile, etc.)
-5. Configuration files and environment setup
-
-Keep your report under 2000 words. Only report facts you verified with tools.`;
-function makeRequirementsAnalystPrompt(task) {
-    return `You are a requirements analyst. Analyze the following task and the repository to identify what needs to be built.
-
-Task: ${task.summary}
-${task.description ? `\nDescription:\n${task.description}` : ""}
-
-IMPORTANT: You MUST use tools to understand the existing codebase before analyzing requirements.
-
-Step 1: Run Glob with pattern "**/*" to see what already exists in the repository.
-Step 2: Read any existing README, docs, or configuration to understand the current state.
-Step 3: Search for any code related to the task requirements using Grep.
-
-After exploring, write a report covering:
-1. Explicit acceptance criteria — what MUST be built based on the description
-2. Implicit requirements — what's assumed but not stated (auth, error handling, etc.)
-3. What already exists vs what needs to be created (based on your file exploration)
-4. Ambiguities that could lead to wrong implementation
-5. Suggested components/modules and which persona should own each
-
-Keep your report under 1500 words.`;
-}
-function makeRiskAssessorPrompt(task) {
-    return `You are a risk assessor. Your job is to search this repository for potential risks and blockers for a development task.
-
-Task: ${task.summary}
-${task.description ? `\nDescription:\n${task.description}` : ""}
-
-IMPORTANT: You MUST use tools to search the codebase. Do NOT guess file paths or make assumptions.
-
-Step 1: Run Glob with pattern "**/*" to see the full repository structure.
-Step 2: Use Grep to search for code related to the task (relevant keywords, APIs, components).
-Step 3: Read files that are likely to be modified or affected by this task.
-Step 4: Search for existing tests (Grep for "test", "spec", "describe", "it(") to find test coverage.
-
-After exploring, write a report covering:
-1. Specific files that will need to be modified (exact paths from your search)
-2. Files with heavy coupling or shared dependencies (imports you found)
-3. Existing tests that will need updating (exact file paths)
-4. Environment, config, or migration requirements
-5. Deployment or infrastructure risks
-
-Keep your report under 1500 words. Only report facts you verified with tools.`;
-}
-/**
- * Run team planning: spawn 3 parallel analyst agents, then synthesize
- * their reports into an enhanced planning prompt for the final planner.
- *
- * Falls back to single-agent planning if anything goes wrong.
- */
-/**
- * Run team analysis: spawn 3 parallel analyst agents once, then return
- * an enhanced prompt with their reports appended. Returns null if all
- * analysts fail (caller should fall back to basePrompt).
- *
- * This runs ONCE before the planner-critic loop — analyst prompts don't
- * include critic feedback, so re-running them on iteration 2+ is waste.
- */
-async function runTeamAnalysis(task, basePrompt, claudePath, model, env, repoPath, taskId, startTime, provider = "anthropic", providerApiKey) {
-    const taskLabel = chalk.cyan(taskId.slice(0, 8));
-    console.log(`${ts()} ${taskLabel} ${chalk.magenta("◆ Team planning")} — running 3 analysts in parallel...`);
-    await postLog(taskId, `${PREFIX} Team planning: running codebase, requirements, and risk analysts in parallel...`);
-    await postProgress(taskId, "reading_repo", Math.round((Date.now() - startTime) / 1000), "Running parallel analysis agents...", 0, 0);
-    const analysisModel = model;
-    const MAX_TEAM_RETRIES = 3;
-    const useCliAnalysts = provider === "anthropic";
-    // Helper: dispatch analyst to Claude CLI or AI SDK based on provider
-    const dispatchAnalyst = (name, prompt) => {
-        if (useCliAnalysts) {
-            return runAnalyst(name, claudePath, analysisModel, prompt, repoPath, env, 900_000, taskId);
-        }
-        if (!providerApiKey) {
-            console.log(`${ts()} ${taskLabel} ${chalk.yellow("⚠")} No API key for ${provider} analysts, skipping ${name}`);
-            return Promise.resolve("");
-        }
-        return runAnalystWithSdk(name, provider, analysisModel, providerApiKey, prompt, repoPath, 900_000, taskId);
-    };
-    let codebaseReport = "";
-    let requirementsReport = "";
-    let riskReport = "";
-    for (let attempt = 1; attempt <= MAX_TEAM_RETRIES; attempt++) {
-        if (attempt > 1) {
-            console.log(`${ts()} ${taskLabel} ${chalk.magenta("◆ Team planning")} — retry ${attempt}/${MAX_TEAM_RETRIES}...`);
-            await postLog(taskId, `${PREFIX} Team analysis retry ${attempt}/${MAX_TEAM_RETRIES}...`);
-        }
-        const [codebaseResult, requirementsResult, riskResult] = await Promise.allSettled([
-            codebaseReport ? Promise.resolve(codebaseReport) : dispatchAnalyst("Codebase", CODEBASE_ANALYST_PROMPT),
-            requirementsReport ? Promise.resolve(requirementsReport) : dispatchAnalyst("Requirements", makeRequirementsAnalystPrompt(task)),
-            riskReport ? Promise.resolve(riskReport) : dispatchAnalyst("Risk", makeRiskAssessorPrompt(task)),
-        ]);
-        if (!codebaseReport && codebaseResult.status === "fulfilled") {
-            codebaseReport = codebaseResult.value;
-        }
-        if (!requirementsReport && requirementsResult.status === "fulfilled") {
-            requirementsReport = requirementsResult.value;
-        }
-        if (!riskReport && riskResult.status === "fulfilled") {
-            riskReport = riskResult.value;
-        }
-        const successCount = [codebaseReport, requirementsReport, riskReport].filter((r) => r.length > 0).length;
-        const analysisElapsed = Math.round((Date.now() - startTime) / 1000);
-        console.log(`${ts()} ${taskLabel} Analysis attempt ${attempt}: ${successCount}/3 reports (${analysisElapsed}s)`);
-        if (successCount > 0) {
-            console.log(`${ts()} ${taskLabel} ${chalk.green("✓")} Analysis complete: ${successCount}/3 reports (${analysisElapsed}s)`);
-            await postLog(taskId, `${PREFIX} Team analysis complete: ${successCount}/3 reports in ${formatElapsed(analysisElapsed)}. Synthesizing plan...`);
-            await postProgress(taskId, "analyzing", analysisElapsed, "Synthesizing analysis reports...", 0, 0);
-            break;
-        }
-        if (attempt === MAX_TEAM_RETRIES) {
-            console.log(`${ts()} ${taskLabel} ${chalk.yellow("⚠")} All analysts failed after ${MAX_TEAM_RETRIES} attempts, falling back to single-agent planning`);
-            await postLog(taskId, `${PREFIX} All analysis agents failed after ${MAX_TEAM_RETRIES} attempts — falling back to single-agent planning`);
-            return null;
-        }
-    }
-    // Build enhanced prompt with analysis reports
-    const sections = [];
-    if (codebaseReport) {
-        sections.push(`## Codebase Analysis (from automated analysis)\n\n${codebaseReport}`);
-    }
-    if (requirementsReport) {
-        sections.push(`## Requirements Analysis\n\n${requirementsReport}`);
-    }
-    if (riskReport) {
-        sections.push(`## Risk Assessment\n\n${riskReport}`);
-    }
-    return (basePrompt +
-        "\n\n" +
-        sections.join("\n\n") +
-        "\n\n" +
-        "Use these analyses to produce a more accurate execution plan.\n" +
-        "Prefer actual file paths discovered in the codebase analysis over guessed paths.");
-}
 /**
  * Run planning for a task with Planner-Critic validation loop.
  *
  * Flow:
  *   1. Fetch planning prompt from cloud API
- *   2. Run Claude CLI to generate plan
- *   3. Parse plan, apply file cap (max 5 files per story)
- *   4. Run critic validation via Claude CLI
- *   5. If critic approves (score >= 80): post validated plan to API
- *   6. If critic rejects: re-run planner with feedback (up to MAX_ITERATIONS)
- *   7. After MAX_ITERATIONS without approval: post best plan if score >= 50 (fallback)
- *   8. If no plan scored >= 50: fail the task
+ *   2. Clone target repo (if available) so planner can explore with tools
+ *   3. Run Claude CLI to generate plan
+ *   4. Parse plan, apply file cap (max 5 files per story)
+ *   5. Run critic validation via Claude CLI
+ *   6. If critic approves (score >= 80): post validated plan to API
+ *   7. If critic rejects: re-run planner with feedback (up to MAX_ITERATIONS)
+ *   8. After MAX_ITERATIONS without approval: post best plan if score >= 50 (fallback)
+ *   9. If no plan scored >= 50: fail the task
  */
 export async function planTask(task, config, credentials) {
     const taskLabel = chalk.cyan(task.id.slice(0, 8));
@@ -816,12 +483,9 @@ export async function planTask(task, config, credentials) {
     const startTime = Date.now();
     // PRD for critic validation: use task description, fall back to summary
     const prd = task.description || task.summary;
-    // Run team analysis ONCE before the planner-critic loop.
-    // Analyst prompts don't include critic feedback, so re-running them
-    // on iteration 2+ wastes compute (they'd produce the same reports).
+    // Clone target repo so the planner can explore with tools
     let repoPath = null;
-    let enhancedBasePrompt = basePrompt;
-    if (config.teamPlanningEnabled && task.githubRepo) {
+    if (task.githubRepo) {
         const scmProvider = task.scmProvider || "github";
         const scmToken = scmProvider === "bitbucket"
             ? config.bitbucketToken
@@ -832,23 +496,11 @@ export async function planTask(task, config, credentials) {
             repoPath = await cloneTargetRepo(task.githubRepo, scmToken, scmProvider, task.id);
         }
         else {
-            console.log(`${ts()} ${taskLabel} ${chalk.yellow("⚠")} No SCM token for ${scmProvider}, skipping team planning`);
-        }
-        if (repoPath) {
-            const analystModel = config.analystModel || cliModel;
-            const analystBackend = isAnthropicPlanning ? "Claude CLI" : `${provider} AI SDK`;
-            console.log(`${ts()} ${taskLabel} Analysts using model: ${chalk.yellow(analystModel)} via ${chalk.dim(analystBackend)} (planner: ${chalk.yellow(cliModel)})`);
-            const analysisResult = await runTeamAnalysis(task, basePrompt, claudePath, analystModel, cleanEnv, repoPath, task.id, startTime, provider, providerApiKey);
-            if (analysisResult) {
-                enhancedBasePrompt = analysisResult;
-            }
-            // else: all analysts failed, fall back to basePrompt
+            console.log(`${ts()} ${taskLabel} ${chalk.yellow("⚠")} No SCM token for ${scmProvider}, planner will run without repo access`);
         }
     }
     // 2. Planner-Critic iteration loop
-    // Use enhancedBasePrompt (with analyst reports) as the base for all iterations.
-    // Critic feedback gets appended on re-plan, but analyst reports are fixed.
-    let currentPrompt = enhancedBasePrompt;
+    let currentPrompt = basePrompt;
     let bestPlan = null;
     let bestScore = 0;
     // Track critic history across iterations for analytics
@@ -870,9 +522,7 @@ export async function planTask(task, config, credentials) {
             let rawOutput;
             try {
                 if (isAnthropicPlanning) {
-                    // Disable tools when analysts already provided repo context
-                    const hasAnalystContext = enhancedBasePrompt !== basePrompt;
-                    rawOutput = await runClaudeCli(claudePath, cliModel, currentPrompt, cleanEnv, task.id, startTime, hasAnalystContext);
+                    rawOutput = await runClaudeCli(claudePath, cliModel, currentPrompt, cleanEnv, task.id, startTime, repoPath || undefined);
                 }
                 else {
                     if (!providerApiKey) {
@@ -997,7 +647,7 @@ export async function planTask(task, config, credentials) {
             // 2f. Rejected — append critic feedback for next iteration
             if (iteration < MAX_ITERATIONS) {
                 const feedback = formatCriticFeedback(criticResult);
-                currentPrompt = enhancedBasePrompt + "\n\n" + feedback;
+                currentPrompt = basePrompt + "\n\n" + feedback;
                 const msg = `${PREFIX} Critic rejected (score: ${criticResult.score}/100, threshold: ${AUTO_APPROVAL_THRESHOLD}). Re-planning with feedback...`;
                 console.log(`${ts()} ${taskLabel} ${chalk.yellow("⚠")} ${msg}`);
                 await postLog(task.id, msg);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@workermill/agent",
-  "version": "0.8.1",
+  "version": "0.8.3",
   "description": "WorkerMill Remote Agent - Run AI workers locally with your Claude Max subscription",
   "type": "module",
   "main": "./dist/index.js",