@xn-intenton-z2a/agentic-lib 7.2.4 → 7.2.6

package/src/actions/agentic-step/tasks/supervise.js

@@ -6,7 +6,7 @@
 // asks the Copilot SDK to choose multiple concurrent actions, then dispatches them.
 
 import * as core from "@actions/core";
-import { existsSync, readFileSync, writeFileSync } from "fs";
+import { existsSync, readFileSync, readdirSync, statSync } from "fs";
 import { runCopilotTask, readOptionalFile, scanDirectory, filterIssues } from "../copilot.js";
 
 /**
@@ -305,6 +305,61 @@ async function gatherContext(octokit, repo, config, t) {
     }
   } catch { /* ignore */ }
 
+  // Check for dedicated test files (not just seed tests)
+  // A dedicated test imports from the source directory (src/lib/) rather than being a seed test
+  let hasDedicatedTests = false;
+  let dedicatedTestFiles = [];
+  try {
+    const testDirs = ["tests", "__tests__"];
+    for (const dir of testDirs) {
+      if (existsSync(dir)) {
+        const testFiles = scanDirectory(dir, [".js", ".ts", ".mjs"], { limit: 20 });
+        for (const tf of testFiles) {
+          // Skip seed test files (main.test.js, web.test.js, behaviour.test.js)
+          if (/^(main|web|behaviour)\.test\.[jt]s$/.test(tf.name)) continue;
+          const content = readFileSync(tf.path, "utf8");
+          // Check if it imports from src/lib/ (mission-specific code)
+          if (/from\s+['"].*src\/lib\//.test(content) || /require\s*\(\s*['"].*src\/lib\//.test(content)) {
+            hasDedicatedTests = true;
+            dedicatedTestFiles.push(tf.name);
+          }
+        }
+      }
+    }
+  } catch { /* ignore */ }
+
+  // W9: Count TODO comments in source directory
+  let sourceTodoCount = 0;
+  try {
+    const sourcePath = config.paths.source?.path || "src/lib/";
+    const sourceDir = sourcePath.endsWith("/") ? sourcePath.slice(0, -1) : sourcePath;
+    const srcRoot = sourceDir.includes("/") ? sourceDir.split("/").slice(0, -1).join("/") || "src" : "src";
+    // Inline recursive TODO counter (avoids circular import with index.js)
+    const countTodos = (dir) => {
+      let n = 0;
+      if (!existsSync(dir)) return 0;
+      try {
+        const entries = readdirSync(dir);
+        for (const entry of entries) {
+          if (entry === "node_modules" || entry.startsWith(".")) continue;
+          const fp = `${dir}/${entry}`;
+          try {
+            const stat = statSync(fp);
+            if (stat.isDirectory()) {
+              n += countTodos(fp);
+            } else if (/\.(js|ts|mjs)$/.test(entry)) {
+              const content = readFileSync(fp, "utf8");
+              const m = content.match(/\bTODO\b/gi);
+              if (m) n += m.length;
+            }
+          } catch { /* skip */ }
+        }
+      } catch { /* skip */ }
+      return n;
+    };
+    sourceTodoCount = countTodos(srcRoot);
+  } catch { /* ignore */ }
+
   return {
     mission,
     recentActivity,
@@ -333,10 +388,13 @@ async function gatherContext(octokit, repo, config, t) {
     cumulativeTransformationCost,
     recentlyClosedSummary,
     sourceExports,
+    hasDedicatedTests,
+    dedicatedTestFiles,
+    sourceTodoCount,
   };
 }
 
-function buildPrompt(ctx, agentInstructions) {
+function buildPrompt(ctx, agentInstructions, config) {
   return [
     "## Instructions",
     agentInstructions,
@@ -368,6 +426,39 @@ function buildPrompt(ctx, agentInstructions) {
           "",
         ]
       : []),
+    `### Test Coverage`,
+    ctx.hasDedicatedTests
+      ? `Dedicated test files: ${ctx.dedicatedTestFiles.join(", ")}`
+      : "**No dedicated test files found.** Only seed tests (main.test.js, web.test.js) exist. Mission-complete requires dedicated tests that import from src/lib/.",
+    "",
+    `### Source TODO Count: ${ctx.sourceTodoCount}`,
+    ctx.sourceTodoCount > 0
+      ? `**${ctx.sourceTodoCount} TODO(s) found in source.** All TODOs must be resolved before mission-complete can be declared.`
+      : "No TODOs found in source — this criterion is met.",
+    "",
+    ...(() => {
+      // W10: Build mission-complete metrics inline for the LLM
+      const thresholds = config?.missionCompleteThresholds || {};
+      const minResolved = thresholds.minResolvedIssues ?? 3;
+      const requireTests = thresholds.requireDedicatedTests ?? true;
+      const maxTodos = thresholds.maxSourceTodos ?? 0;
+      const resolvedCount = ctx.recentlyClosedSummary.filter((s) => s.includes("RESOLVED")).length;
+      const rows = [
+        `### Mission-Complete Metrics`,
+        "| Metric | Value | Target | Status |",
+        "|--------|-------|--------|--------|",
+        `| Open issues | ${ctx.issuesSummary.length} | 0 | ${ctx.issuesSummary.length === 0 ? "MET" : "NOT MET"} |`,
+        `| Open PRs | ${ctx.prsSummary.length} | 0 | ${ctx.prsSummary.length === 0 ? "MET" : "NOT MET"} |`,
+        `| Issues resolved (RESOLVED) | ${resolvedCount} | >= ${minResolved} | ${resolvedCount >= minResolved ? "MET" : "NOT MET"} |`,
+        `| Dedicated test files | ${ctx.hasDedicatedTests ? "YES" : "NO"} | ${requireTests ? "YES" : "—"} | ${!requireTests || ctx.hasDedicatedTests ? "MET" : "NOT MET"} |`,
+        `| Source TODO count | ${ctx.sourceTodoCount} | <= ${maxTodos} | ${ctx.sourceTodoCount <= maxTodos ? "MET" : "NOT MET"} |`,
+        `| Budget used | ${ctx.cumulativeTransformationCost}/${ctx.transformationBudget} | < ${ctx.transformationBudget || "unlimited"} | ${ctx.transformationBudget > 0 && ctx.cumulativeTransformationCost >= ctx.transformationBudget ? "EXHAUSTED" : "OK"} |`,
+        "",
+        "**All metrics must show MET/OK for mission-complete to be declared.**",
+        "",
+      ];
+      return rows;
+    })(),
     `### Recent Workflow Runs`,
     ctx.workflowsSummary.join("\n") || "none",
     "",
@@ -419,7 +510,7 @@ function buildPrompt(ctx, agentInstructions) {
         ]
       : []),
     ...(ctx.transformationBudget > 0
-      ? [`### Transformation Budget: ${ctx.cumulativeTransformationCost}/${ctx.transformationBudget} used (${Math.max(0, ctx.transformationBudget - ctx.cumulativeTransformationCost)} remaining)`, ""]
+      ? [`### Transformation Budget: ${ctx.cumulativeTransformationCost}/${ctx.transformationBudget} used (${Math.max(0, ctx.transformationBudget - ctx.cumulativeTransformationCost)} remaining)`, "Note: instability transforms (infrastructure fixes) do not count against this budget.", ""]
       : []),
     `### Issue Limits`,
     `Feature development WIP limit: ${ctx.featureIssuesWipLimit}`,
@@ -444,10 +535,6 @@ function buildPrompt(ctx, agentInstructions) {
     "### Communication",
     "- `respond:discussions | message: <text> | discussion-url: <url>` — Reply via discussions bot",
     "",
-    "### Mission Lifecycle",
-    "- `mission-complete | reason: <text>` — Declare mission accomplished. Writes MISSION_COMPLETE.md and sets schedule to off. Use when: all acceptance criteria in MISSION.md are satisfied, tests pass, and recently-closed issues confirm resolution.",
-    "- `mission-failed | reason: <text>` — Declare mission failed. Writes MISSION_FAILED.md and sets schedule to off. Use when: transformation budget is exhausted with no progress, pipeline is stuck in a loop, or the mission is unachievable.",
-    "",
     "### Schedule Control",
     "- `set-schedule:<frequency>` — Change supervisor schedule (off, weekly, daily, hourly, continuous). Use `set-schedule:weekly` when mission is substantially complete, `set-schedule:continuous` to ramp up.",
     "",
@@ -605,152 +692,11 @@ async function executeRespondDiscussions(octokit, repo, params, ctx) {
   return "skipped:respond-no-message";
 }
 
-async function executeMissionComplete(octokit, repo, params, ctx) {
-  const reason = params.reason || "All acceptance criteria satisfied";
-  const signal = [
-    "# Mission Complete",
-    "",
-    `- **Timestamp:** ${new Date().toISOString()}`,
-    `- **Detected by:** supervisor`,
-    `- **Reason:** ${reason}`,
-    "",
-    "This file was created automatically. To restart transformations, delete this file or run `npx @xn-intenton-z2a/agentic-lib init --reseed`.",
-  ].join("\n");
-  writeFileSync("MISSION_COMPLETE.md", signal);
-  core.info(`Mission complete signal written: ${reason}`);
-
-  // Persist MISSION_COMPLETE.md to the repository via Contents API so it survives across runs
-  try {
-    const contentBase64 = Buffer.from(signal).toString("base64");
-    // Check if file already exists (to get its SHA for updates)
-    let existingSha;
-    try {
-      const { data } = await octokit.rest.repos.getContent({ ...repo, path: "MISSION_COMPLETE.md", ref: "main" });
-      existingSha = data.sha;
-    } catch {
-      // File doesn't exist yet — that's fine
-    }
-    await octokit.rest.repos.createOrUpdateFileContents({
-      ...repo,
-      path: "MISSION_COMPLETE.md",
-      message: "mission-complete: " + reason.substring(0, 72),
-      content: contentBase64,
-      branch: "main",
-      ...(existingSha ? { sha: existingSha } : {}),
-    });
-    core.info("MISSION_COMPLETE.md committed to main via Contents API");
-  } catch (err) {
-    core.warning(`Could not commit MISSION_COMPLETE.md to repo: ${err.message}`);
-  }
-
-  if (process.env.GITHUB_REPOSITORY !== "xn-intenton-z2a/agentic-lib") {
-    // Only turn off schedule if it's not already off or in maintenance mode
-    let currentSupervisor = "";
-    try {
-      const tomlContent = readFileSync("agentic-lib.toml", "utf8");
-      const match = tomlContent.match(/^\s*supervisor\s*=\s*"([^"]*)"/m);
-      if (match) currentSupervisor = match[1];
-    } catch { /* ignore */ }
-
-    if (currentSupervisor === "off" || currentSupervisor === "maintenance") {
-      core.info(`Schedule already "${currentSupervisor}" — not changing on mission-complete`);
-    } else {
-      try {
-        await octokit.rest.actions.createWorkflowDispatch({
-          ...repo,
-          workflow_id: "agentic-lib-schedule.yml",
-          ref: "main",
-          inputs: { frequency: "off" },
-        });
-      } catch (err) {
-        core.warning(`Could not set schedule to off: ${err.message}`);
-      }
-    }
-
-    // Announce mission complete via bot
-    const websiteUrl = getWebsiteUrl(repo);
-    const discussionUrl = ctx?.activeDiscussionUrl || "";
-    await dispatchBot(octokit, repo, `Mission complete! ${reason}\n\nWebsite: ${websiteUrl}`, discussionUrl);
-  }
-  return `mission-complete:${reason.substring(0, 100)}`;
-}
-
-async function executeMissionFailed(octokit, repo, params, ctx) {
-  const reason = params.reason || "Mission could not be completed";
-  const signal = [
-    "# Mission Failed",
-    "",
-    `- **Timestamp:** ${new Date().toISOString()}`,
-    `- **Detected by:** supervisor`,
-    `- **Reason:** ${reason}`,
-    "",
-    "This file was created automatically. To restart, delete this file and run `npx @xn-intenton-z2a/agentic-lib init --reseed`.",
-  ].join("\n");
-  writeFileSync("MISSION_FAILED.md", signal);
-  core.info(`Mission failed signal written: ${reason}`);
-
-  // Persist MISSION_FAILED.md to the repository via Contents API so it survives across runs
-  try {
-    const contentBase64 = Buffer.from(signal).toString("base64");
-    let existingSha;
-    try {
-      const { data } = await octokit.rest.repos.getContent({ ...repo, path: "MISSION_FAILED.md", ref: "main" });
-      existingSha = data.sha;
-    } catch {
-      // File doesn't exist yet — that's fine
-    }
-    await octokit.rest.repos.createOrUpdateFileContents({
-      ...repo,
-      path: "MISSION_FAILED.md",
-      message: "mission-failed: " + reason.substring(0, 72),
-      content: contentBase64,
-      branch: "main",
-      ...(existingSha ? { sha: existingSha } : {}),
-    });
-    core.info("MISSION_FAILED.md committed to main via Contents API");
-  } catch (err) {
-    core.warning(`Could not commit MISSION_FAILED.md to repo: ${err.message}`);
-  }
-
-  if (process.env.GITHUB_REPOSITORY !== "xn-intenton-z2a/agentic-lib") {
-    // Only turn off schedule if it's not already off or in maintenance mode
-    let currentSupervisor = "";
-    try {
-      const tomlContent = readFileSync("agentic-lib.toml", "utf8");
-      const match = tomlContent.match(/^\s*supervisor\s*=\s*"([^"]*)"/m);
-      if (match) currentSupervisor = match[1];
-    } catch { /* ignore */ }
-
-    if (currentSupervisor === "off" || currentSupervisor === "maintenance") {
-      core.info(`Schedule already "${currentSupervisor}" — not changing on mission-failed`);
-    } else {
-      try {
-        await octokit.rest.actions.createWorkflowDispatch({
-          ...repo,
-          workflow_id: "agentic-lib-schedule.yml",
-          ref: "main",
-          inputs: { frequency: "off" },
-        });
-      } catch (err) {
-        core.warning(`Could not set schedule to off: ${err.message}`);
-      }
-    }
-
-    // Announce mission failed via bot
-    const websiteUrl = getWebsiteUrl(repo);
-    const discussionUrl = ctx?.activeDiscussionUrl || "";
-    await dispatchBot(octokit, repo, `Mission failed. ${reason}\n\nWebsite: ${websiteUrl}`, discussionUrl);
-  }
-  return `mission-failed:${reason.substring(0, 100)}`;
-}
-
 const ACTION_HANDLERS = {
   "github:create-issue": executeCreateIssue,
   "github:label-issue": executeLabelIssue,
   "github:close-issue": executeCloseIssue,
   "respond:discussions": executeRespondDiscussions,
-  "mission-complete": executeMissionComplete,
-  "mission-failed": executeMissionFailed,
 };
 
 async function executeSetSchedule(octokit, repo, frequency) {
@@ -778,7 +724,7 @@ async function executeAction(octokit, repo, action, params, ctx) {
   if (action === "nop") return "nop";
   const handler = ACTION_HANDLERS[action];
   if (handler) return handler(octokit, repo, params, ctx);
-  core.warning(`Unknown action: ${action}`);
+  core.debug(`Ignoring unrecognised action: ${action}`);
   return `unknown:${action}`;
 }
 
@@ -814,7 +760,7 @@ export async function supervise(context) {
 
   // --- LLM decision ---
   const agentInstructions = instructions || "You are the supervisor. Decide what actions to take.";
-  const prompt = buildPrompt(ctx, agentInstructions);
+  const prompt = buildPrompt(ctx, agentInstructions, config);
 
   const { content, tokensUsed, inputTokens, outputTokens, cost } = await runCopilotTask({
     model,
@@ -845,28 +791,8 @@ export async function supervise(context) {
 
   // --- Deterministic lifecycle posts (after LLM) ---
 
-  // Strategy A: Deterministic mission-complete fallback
-  // If the LLM didn't choose mission-complete but conditions are clearly met, auto-execute it.
-  // Skip in maintenance mode — maintenance keeps running regardless of mission status.
-  if (!ctx.missionComplete && !ctx.missionFailed && config.supervisor !== "maintenance") {
-    const llmChoseMissionComplete = results.some((r) => r.startsWith("mission-complete:"));
-    if (!llmChoseMissionComplete) {
-      const resolvedCount = ctx.recentlyClosedSummary.filter((s) => s.includes("RESOLVED")).length;
-      const hasNoOpenIssues = ctx.issuesSummary.length === 0;
-      const hasNoOpenPRs = ctx.prsSummary.length === 0;
-      if (hasNoOpenIssues && hasNoOpenPRs && resolvedCount >= 1) {
-        core.info(`Deterministic mission-complete: 0 open issues, 0 open PRs, ${resolvedCount} recently resolved — LLM did not detect completion`);
-        try {
-          const autoResult = await executeMissionComplete(octokit, repo,
-            { reason: `All acceptance criteria satisfied (${resolvedCount} issues resolved, 0 open issues, 0 open PRs)` },
-            ctx);
-          results.push(autoResult);
-        } catch (err) {
-          core.warning(`Deterministic mission-complete failed: ${err.message}`);
-        }
-      }
-    }
-  }
+  // W12: Mission-complete/failed evaluation moved to the director task.
+  // The supervisor no longer declares mission-complete or mission-failed.
 
   // Step 3: Auto-respond when a message referral is present
   // If the workflow was triggered with a message (from bot's request-supervisor),
@@ -888,14 +814,8 @@ export async function supervise(context) {
 
   // Build changes list from executed actions
   const changes = results
-    .filter((r) => r.startsWith("created-issue:") || r.startsWith("mission-complete:") || r.startsWith("mission-failed:"))
-    .map((r) => {
-      if (r.startsWith("created-issue:")) return { action: "created-issue", file: r.replace("created-issue:", ""), sizeInfo: "" };
-      if (r.startsWith("mission-complete:")) return { action: "mission-complete", file: "MISSION_COMPLETE.md", sizeInfo: r.replace("mission-complete:", "") };
-      if (r.startsWith("mission-failed:")) return { action: "mission-failed", file: "MISSION_FAILED.md", sizeInfo: r.replace("mission-failed:", "") };
-      return null;
-    })
-    .filter(Boolean);
+    .filter((r) => r.startsWith("created-issue:"))
+    .map((r) => ({ action: "created-issue", file: r.replace("created-issue:", ""), sizeInfo: "" }));
 
   return {
     outcome: actions.length === 0 ? "nop" : `supervised:${actions.length}-actions`,

package/src/agents/agent-director.md

@@ -0,0 +1,58 @@
+You are the director of an autonomous coding repository. Your sole responsibility is evaluating whether the mission is complete, failed, or in progress.
+
+## Your Role
+
+You do NOT dispatch workflows, create issues, or manage the schedule. That is the supervisor's job. You ONLY assess mission status and produce a structured evaluation.
+
+## Input
+
+You receive:
+1. **MISSION.md** — the acceptance criteria
+2. **Mission-Complete Metrics** — a table of mechanical checks (open issues, PRs, resolved count, test coverage, TODO count, budget)
+3. **Metric based mission complete assessment** — a pre-computed advisory from the mechanical check
+4. **Source Exports** — functions exported from source files
+5. **Recently Closed Issues** — issues resolved since init
+6. **Recent Activity** — the latest entries from the activity log
+
+## Decision Framework
+
+### Mission Complete
+Declare `mission-complete` when ALL of the following are true:
+1. Every row in the Mission-Complete Metrics table shows **MET** or **OK**
+2. The Source Exports demonstrate that all functions required by MISSION.md are implemented
+3. The Recently Closed Issues confirm that acceptance criteria have been addressed
+4. No TODOs remain in source code
+5. Dedicated test files exist (not just seed tests)
+
+### Mission Failed
+Declare `mission-failed` when ANY of the following are true:
+1. Transformation budget is EXHAUSTED and acceptance criteria are still unmet
+2. The last 3+ transforms produced no meaningful code changes
+3. The pipeline is stuck in a loop (same issues created and closed repeatedly)
+
+### Gap Analysis (most common output)
+When the mission is neither complete nor failed, produce a detailed gap analysis:
+- What has been achieved so far
+- What specific gaps remain between the current state and mission-complete
+- Which metrics are NOT MET and what needs to happen to satisfy them
+- Prioritised list of what should be done next
+
+## Output Format
+
+Respond with EXACTLY this structure:
+
+```
+[DECISION]
+mission-complete | mission-failed | in-progress
+[/DECISION]
+[REASON]
+One-line summary of the decision.
+[/REASON]
+[ANALYSIS]
+Detailed gap analysis or completion summary. Include:
+- Metrics status (which are MET, which are NOT MET)
+- What has been achieved
+- What remains (if in-progress)
+- Recommended next actions (if in-progress)
+[/ANALYSIS]
+```

package/src/agents/agent-supervisor.md

@@ -1,18 +1,24 @@
 You are the supervisor of an autonomous coding repository. Your job is to advance the mission by strategically choosing which workflows to dispatch and which GitHub actions to take.
 
-## MANDATORY FIRST CHECK: Is the Mission Already Complete?
+**Important:** You do NOT evaluate mission-complete or mission-failed. That is the director's exclusive responsibility. Focus on advancing the mission through strategic action.
 
-**Before choosing ANY action, evaluate this:**
+## MANDATORY FIRST CHECK: What Needs to Happen Next?
 
-1. Are there 0 open issues?
-2. Were 2+ recently-closed issues "closed by review as RESOLVED"?
-3. Do the Source Exports show the functions required by MISSION.md?
+**Before choosing ANY action, check the Mission-Complete Metrics table in the prompt.**
 
-If ALL three are true → the mission is done. Choose `mission-complete | reason: <summary>`. Do NOT create another issue for work that is already implemented and reviewed.
+Look at which metrics are NOT MET — these tell you what gaps remain:
+1. Open issues > 0 → close resolved issues or wait for review
+2. Open PRs > 0 → merge or close stale PRs
+3. Issues resolved < threshold → create and resolve more issues
+4. Dedicated test files = NO → create an issue requesting dedicated tests
+5. Source TODO count > 0 → create an issue to resolve TODOs
+6. Budget near exhaustion → be strategic with remaining transforms
+
+If all metrics show MET/OK, use `nop` — the director will handle the rest.
 
 ## Priority Order
 
-1. **Always strive for mission complete** — every action you take should aim to finish the mission. If the code is already complete (see Source Exports and Recently Closed Issues), declare `mission-complete` immediately. Otherwise, create one comprehensive issue that targets the entire mission (all acceptance criteria, tests, website, docs, README). Only create a second issue if the first transform couldn't complete everything, and scope it to the remaining work. Do not create issues just to fill a quota.
+1. **Always strive to close gaps** — every action you take should aim to satisfy the remaining NOT MET metrics. If the code is already complete (see Source Exports and Recently Closed Issues), use `nop` and let the director evaluate. Otherwise, create one comprehensive issue that targets the entire mission (all acceptance criteria, tests, website, docs, README). Only create a second issue if the first transform couldn't complete everything, and scope it to the remaining work. Do not create issues just to fill a quota.
 2. **Dispatch transform when ready issues exist** — transform is where code gets written. Always prefer it over maintain when there are open issues with the `ready` label.
 3. **Dispatch review after transform** — when recent workflow runs show a transform completion, dispatch review to close resolved issues and add `ready` labels to new issues. This keeps the pipeline flowing.
 4. **Fix failing PRs** — dispatch fix-code for any PR with failing checks (include pr-number).
@@ -36,10 +42,8 @@ If ALL three are true → the mission is done. Choose `mission-complete | reason
 - **github:label-issue** — When an issue needs better categorisation for prioritisation.
 - **github:close-issue** — When an issue is clearly resolved or no longer relevant.
 - **respond:discussions** — When replying to a user request that came through the discussions bot. Include the discussion URL and a clear message.
-- **set-schedule:\<frequency\>** — Change the workflow schedule. Use `weekly` when mission is substantially achieved, `continuous` to ramp up for active development.
-- **mission-complete** — When all MISSION.md acceptance criteria are verified as satisfied. Review the Recently Closed Issues — if the last 2+ issues were closed by review as RESOLVED, 0 open issues remain, and the acceptance criteria in MISSION.md match the implemented code, declare mission complete. This writes MISSION_COMPLETE.md and sets the schedule to off. Always include a reason summarising what was achieved.
-- **mission-failed** — When the mission cannot be completed. Use when: transformation budget is exhausted with acceptance criteria still unmet, the pipeline is stuck in a create-close loop with no code changes, or 3+ consecutive transforms failed to produce working code. This writes MISSION_FAILED.md and sets the schedule to off. Always include a reason explaining what went wrong.
-- **nop** — When everything is running optimally: transform is active, issues are flowing, no failures.
+- **set-schedule:\<frequency\>** — Change the workflow schedule. Use `weekly` when activity is low, `continuous` to ramp up for active development.
+- **nop** — When everything is running optimally: transform is active, issues are flowing, no failures. Also use when all metrics are MET — let the director handle the evaluation.
 
 ## Stale Issue Detection
 
@@ -52,43 +56,11 @@ When recent workflow runs show an init completion, the repository has a fresh or
 Dispatch the discussions bot to announce the new mission to the community.
 Include the website URL in the announcement — the site is at `https://<owner>.github.io/<repo>/` and runs the library.
 
-### Mission Accomplished (bounded missions)
-When ALL of the following conditions are met, the mission is accomplished:
-1. All open issues are closed (check Recently Closed Issues — if the last 2+ were closed by review as RESOLVED, this is strong evidence)
-2. Tests pass (CI gates commits, so this is usually the case)
-3. The MISSION.md acceptance criteria are all satisfied (verify each criterion against the Recently Closed Issues and Recent Activity)
-4. Do not create an issue if a similar issue was recently closed as resolved — check the Recently Closed Issues section
-
-When all conditions are met, use the `mission-complete` action:
-1. `mission-complete | reason: <summary of what was achieved>` — this writes MISSION_COMPLETE.md and sets the schedule to off
-2. `dispatch:agentic-lib-bot` — announce mission accomplished in the discussions thread. Include the website URL (`https://<owner>.github.io/<repo>/`) where users can see the finished product.
-
-Do NOT create another issue when the mission is already accomplished. If the Recently Closed Issues show 2+ issues closed by review as RESOLVED and 0 open issues remain, the mission is done.
-
 ### Ongoing Missions
 If MISSION.md explicitly says "do not set schedule to off" or "ongoing mission", the mission never completes.
 Instead, when activity is healthy, use `set-schedule:weekly` or `set-schedule:daily` to keep the pipeline running.
 Never use `set-schedule:off` for ongoing missions.
 
-### Mission Substantially Complete (bounded, but minor gaps)
-When the transform agent has implemented all major features but minor polish remains
-(e.g. missing README examples, incomplete edge case coverage):
-1. `dispatch:agentic-lib-bot` — announce near-completion in the discussions thread
-2. `set-schedule:weekly` — reduce to weekly maintenance check-ins
-3. Check that `docs/` contains evidence of the library working before declaring done
-
-### Mission Failed
-When the mission cannot be completed, use the `mission-failed` action. Indicators of failure:
-1. **Budget exhausted** — Transformation Budget shows usage at or near capacity with acceptance criteria still unmet
-2. **Pipeline stuck** — 3+ consecutive supervisor cycles created issues that were immediately closed by review as RESOLVED, but the acceptance criteria are NOT actually met (false positives in review)
-3. **No progress** — the last 3+ transforms produced no code changes (all nop outcomes) and acceptance criteria remain unmet
-4. **Repeated failures** — transforms keep producing code that fails tests, and fix-code cannot resolve the failures
-5. **Consuming budget without results** — transformation budget is being spent but the codebase is not converging toward the acceptance criteria
-
-When declaring mission failed:
-1. `mission-failed | reason: <what went wrong and what was achieved>` — this writes MISSION_FAILED.md and sets the schedule to off
-2. `dispatch:agentic-lib-bot` — announce the failure in the discussions thread with details of what was accomplished and what remains
-
 ## Prerequisites
 
 - The `set-schedule` action requires a `WORKFLOW_TOKEN` secret (classic PAT with `workflow` scope) to push workflow file changes to main.
@@ -97,13 +69,13 @@ When declaring mission failed:
 
 Check the Recent Activity log and Recently Closed Issues for patterns:
 
-**Mission complete signals:**
-- If the last 2+ issues were closed by review as RESOLVED, AND 0 open issues remain, the mission is likely accomplished. Verify against MISSION.md acceptance criteria, then use `mission-complete`.
-- If the last 2+ workflow runs produced no transform commits (only maintain-only or nop outcomes), AND all open issues are closed, follow the "Mission Accomplished" protocol.
+**All metrics MET signals:**
+- If all rows in the Mission-Complete Metrics table show MET/OK, use `nop` — the director will evaluate mission-complete.
+- If the last 2+ workflow runs produced no transform commits (only maintain-only or nop outcomes), AND all open issues are closed, use `nop`.
 
-**Mission failed signals:**
-- If the Transformation Budget shows usage near capacity (e.g. 28/32) and acceptance criteria are still unmet, the mission is failing. Use `mission-failed`.
-- If the last 3+ cycles show the pattern: create issue → review closes as resolved → no transform → create identical issue, the pipeline is stuck. Check if acceptance criteria are truly met (use `mission-complete`) or if review is wrong (create a more specific issue). If neither works, use `mission-failed`.
+**Budget exhaustion signals:**
+- If the Transformation Budget shows usage near capacity (e.g. 28/32) and acceptance criteria are still unmet, be strategic with remaining budget. Create highly-targeted issues that address the most critical gaps.
+- If the last 3+ cycles show the pattern: create issue → review closes as resolved → no transform → create identical issue, the pipeline is stuck. Check if acceptance criteria are truly met (metrics will reflect this) or if review is wrong (create a more specific issue).
 - Look for `transform: nop` or `transform: transformed` patterns in the activity log to distinguish productive iterations from idle ones.
 
 **Dedup deadlock recovery:**
@@ -115,7 +87,7 @@ Check the Recent Activity log for discussion bot referrals (lines containing `di
 
 Also check for notable progress worth reporting:
 - Mission milestones achieved (all core functions implemented, all tests passing)
-- Schedule changes (mission accomplished, throttling down)
+- Schedule changes (throttling down)
 - Significant code changes (large PRs merged, new features completed)
 - Website first deployed or significantly updated (include the URL: `https://<owner>.github.io/<repo>/`)
 

package/src/seeds/zero-behaviour.test.js

@@ -3,12 +3,12 @@
 import { test, expect } from "@playwright/test";
 
 test("homepage returns 200 and renders", async ({ page }) => {
-  const response = await page.goto("/");
+  const response = await page.goto("/", { waitUntil: "networkidle" });
   expect(response.status()).toBe(200);
 
-  await expect(page.locator("#lib-name")).toBeVisible();
-  await expect(page.locator("#lib-version")).toBeVisible();
-  await expect(page.locator("#demo-output")).toBeVisible();
+  await expect(page.locator("#lib-name")).toBeVisible({ timeout: 10000 });
+  await expect(page.locator("#lib-version")).toBeVisible({ timeout: 10000 });
+  await expect(page.locator("#demo-output")).toBeVisible({ timeout: 10000 });
 
   await page.screenshot({ path: "SCREENSHOT_INDEX.png", fullPage: true });
 });

package/src/seeds/zero-package.json

@@ -17,7 +17,7 @@
   "author": "",
   "license": "MIT",
   "dependencies": {
-    "@xn-intenton-z2a/agentic-lib": "^7.2.4"
+    "@xn-intenton-z2a/agentic-lib": "^7.2.6"
   },
   "devDependencies": {
     "@playwright/test": "^1.58.0",

package/src/seeds/zero-playwright.config.js

@@ -5,6 +5,7 @@ import { defineConfig } from "@playwright/test";
 export default defineConfig({
   testDir: "tests/behaviour",
   timeout: 30000,
+  retries: 2,
   use: {
     baseURL: "http://localhost:3000",
   },