@ionivetech/mugiwara 0.6.0 → 0.6.1

package/.claude-plugin/marketplace.json

@@ -5,13 +5,13 @@
   },
   "metadata": {
     "description": "The Straw Hat crew for AI agents",
-    "version": "0.6.0"
+    "version": "0.6.1"
   },
   "plugins": [
     {
       "name": "mugiwara",
       "description": "The Straw Hat crew of AI agents and skills: brainstorm, plan, execute, checkpoint, quality, gates, review, security, healing.",
-      "version": "0.6.0",
+      "version": "0.6.1",
       "source": "./"
     }
   ]

package/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "mugiwara",
   "displayName": "Mugiwara",
-  "version": "0.6.0",
+  "version": "0.6.1",
   "description": "The Straw Hat crew of AI agents and skills: brainstorm, plan, execute, checkpoint, quality, gates, review, security, healing.",
   "author": {
     "name": "ionivetech"

package/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "mugiwara",
-  "version": "0.6.0",
+  "version": "0.6.1",
   "description": "The Straw Hat crew of AI agents and skills: brainstorm, plan, execute, checkpoint, quality, gates, review, security, healing.",
   "author": {
     "name": "ionivetech"

package/.cursor-plugin/plugin.json

@@ -2,7 +2,7 @@
   "name": "mugiwara",
   "displayName": "Mugiwara",
   "description": "The Straw Hat crew of AI agents and skills: brainstorm, plan, execute, checkpoint, quality, gates, review, security, healing.",
-  "version": "0.6.0",
+  "version": "0.6.1",
   "author": {
     "name": "ionivetech"
   },

package/.kimi-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "mugiwara",
-  "version": "0.6.0",
+  "version": "0.6.1",
   "description": "The Straw Hat crew of AI agents and skills: brainstorm, plan, execute, checkpoint, quality, gates, review, security, healing.",
   "author": {
     "name": "ionivetech"

package/.opencode/mugiwara-helpers.mjs

@@ -88,3 +88,27 @@ export function applyModeChange(mode, { projectDir = process.cwd(), home = homed
   mkdirSync(dirname(log), { recursive: true });
   appendFileSync(log, `| ${new Date().toISOString()} | mode flip | guided/semi/auto -> ${mode} | user |\n`);
 }
+
+export const DEFAULT_CONFIG_LINES = [
+  'mode=guided',
+  'branch=feature/{type}-{issue}-{slug}',
+  'commit=conventional',
+  'base=main',
+  'coverage_new=90',
+  'coverage_modified=80',
+  'review_depth=full',
+  'quality_depth=full',
+];
+
+// Idempotent: writes the full default config only when .mugiwara/config is
+// absent (a fresh repo's first use). Never overwrites an existing config —
+// mode=guided is the safe default; the user's later edits win.
+export function ensureDefaultConfig({ projectDir = process.cwd() } = {}) {
+  const dir = join(projectDir, '.mugiwara');
+  const file = join(dir, 'config');
+  if (existsSync(file)) return false;
+  assertNotSymlink(file);
+  mkdirSync(dir, { recursive: true });
+  writeFileSync(file, DEFAULT_CONFIG_LINES.join('\n') + '\n');
+  return true;
+}

package/.opencode/plugins/mugiwara.mjs

@@ -14,7 +14,7 @@
 import { existsSync, readdirSync, readFileSync } from 'node:fs';
 import { dirname, join } from 'node:path';
 import { fileURLToPath } from 'node:url';
-import { readMode, parseModeChange, applyModeChange } from '../mugiwara-helpers.mjs';
+import { readMode, parseModeChange, applyModeChange, ensureDefaultConfig } from '../mugiwara-helpers.mjs';
 
 const __dirname = dirname(fileURLToPath(import.meta.url));
 
@@ -82,7 +82,12 @@ function readAgents() {
       continue;
     }
     if (!parsed.data.description || !parsed.body) continue;
-    agents[name] = { description: parsed.data.description, mode: 'all', prompt: parsed.body };
+    const internal = parsed.data.internal === 'true';
+    agents[name] = {
+      description: internal ? `[INTERNAL] ${parsed.data.description}` : parsed.data.description,
+      mode: internal ? 'subagent' : 'all',
+      prompt: parsed.body,
+    };
     if (CREW[name]) agents[name] = { ...agents[name], ...CREW[name] };
     const perm = permissionFromScope(parsed.data['write-scope']);
     if (perm) agents[name].permission = perm;
@@ -94,6 +99,9 @@ export default async () => ({
   dispose: () => {},
 
   config: (config) => {
+    // only seed .mugiwara/config when the cwd looks like a project — a global
+    // install must not create .mugiwara/ in an arbitrary non-project dir.
+    if (existsSync(join(process.cwd(), '.git'))) ensureDefaultConfig();
     config.skills = config.skills || {};
     config.skills.paths = config.skills.paths || [];
     if (!config.skills.paths.includes(skillsDir)) config.skills.paths.push(skillsDir);

package/AGENTS.md

@@ -173,7 +173,7 @@ Conventional Commits: `feat:`, `fix:`, `refactor:`, `docs:`, `chore:`.
 
 ```
 content/skills/    — 26 skill dirs, each with SKILL.md + optional references/
-content/agents/    — 14 agent .md files (11 user-facing + 3 internal)
+content/agents/    — 15 agent .md files (12 user-facing + 3 internal)
 references/         — shared reference files (definition-of-done, source-grounding, etc.)
 docs/               — user-facing documentation
 scripts/            — validation + tooling scripts

package/README.md

@@ -4,10 +4,12 @@
 [![npm downloads](https://img.shields.io/npm/dm/@ionivetech%2fmugiwara)](https://www.npmjs.com/package/@ionivetech/mugiwara)
 [![License: MIT](https://img.shields.io/github/license/ionivetech/mugiwara)](https://github.com/ionivetech/mugiwara/blob/main/LICENSE)
 
-**Ship quality code, not just fast code.** Mugiwara gives your AI agent a
-governed engineering team — 12 specialists who plan, build, audit, review, and
-heal — with evidence at every step. No runtime, no API keys, no servers. Just
-markdown your agent already knows how to read.
+**Your agent writes the code. Mugiwara proves it.**
+
+A governed engineering crew for your AI agent — evidence at every step, and a
+process that sizes itself to the work. A typo costs nothing. An auth migration
+gets all nine waves. No runtime, no API keys, no servers. Just markdown your
+agent already knows how to read.
 
 Works on Claude Code, opencode, Copilot, Gemini, and 8 more platforms.
 
@@ -21,43 +23,63 @@ every agent, every skill, every rule is static markdown.
 
 → [Full pitch: why mugiwara vs just asking your agent](docs/concepts/comparison.md)
 
-## What Mugiwara does
+## See the evidence
 
-### 9-wave pipeline
+A closed mission leaves a report you can actually read. This is the exact
+format `scripts/mission-report.sh` produces:
 
-```mermaid
-flowchart TB
-    L0["Luffy<br>Triage"] --> L1["Usopp<br>Brainstorm"] --> L2["Nami<br>Plan"] --> L3["Zoro<br>Execute"] --> L4["Chopper<br>Audit"]
-    L4 --> L5["Sanji<br>Quality"] --> L6["Franky<br>Gates"]
-    L6 --> L7R["Robin<br>Review"]
-    L6 --> L7J["Jinbe<br>Security"]
-    L7R --> L8["Brook<br>Heal"]
-    L7J --> L8
-    L8 --> L9["Luffy<br>Closure"]
-    L8 -. "heal ≤3 cycles" .-> L4
-```
+    # Mission: invitation-accepted-flow . 2026-08-11
 
-→ [Full pipeline details](docs/concepts/workflow.md)
+    **Lane** full . **Mode** guided . **Actor** farid . **Branch** feature/MKR-412
 
-### All features
+    ## What changed
+
+    11 files, +340 LOC
+    Sensitive paths: src/auth/
+
+    ## Waves
+
+    | Wave | Artifact | Verdict |
+    |------|----------|---------|
+    | Execute (Wave 3) | `01-execution.md` | PASS |
+    | Checkpoint (Wave 4) | `02-audit.md` | PASS |
+    | Quality (Wave 5) | `03-quality.md` | PASS |
+    | Gates (Wave 6) | `04-gates.md` | PASS |
+    | Healing (Wave 8) | `05-healing.md` | PASS |
+    | Closure (Wave 9) | `06-closure.md` | GO |
+
+    ## Review & blockers
 
-| Feature                     | What you get                                                                                    |
-| --------------------------- | ----------------------------------------------------------------------------------------------- |
-| **Lane sizing**             | Work auto-sized from `git diff`. Typo = instant fix. Auth migration = full pipeline.            |
-| **Team initiatives**        | Sub-missions with assignee + branch, shared plan doc, status tracking per sub-mission.          |
-| **Sonar-style quality**     | Duplication %, complexity, maintainability rating (A–E), code attributes, per-condition gate.   |
-| **STRIDE + OWASP security** | Threat modeling, secret scan, injection check, SCA license, hotspot review.                     |
-| **Self-healing**            | Brook reads all failures at once, fixes root causes, re-runs verification. ≤3 cycles.           |
-| **3 autonomy modes**        | guided (ask) / semi (auto branch+commit) / auto (hands-off). Flip mid-session.                  |
-| **Configurable depth**      | `review_depth` + `quality_depth`: full / standard / quick. Set per project.                     |
-| **Resume from anywhere**    | Session lost? Rebuilds from `.mugiwara/state.json`. Continues, never restarts.                  |
-| **Onboarding wizard**       | 10 questions — project type, team size, mode, depth. `/mugiwara onboard`.                       |
-| **Evidence trail**          | `.mugiwara/` workspace: plans, audit reports, quality reports, review findings, blocker ledger. |
-| **12 platforms**            | Claude Code, opencode, Copilot, Gemini, Codex, Cursor, Kimi, Pi, Antigravity + CLI.             |
-| **Cost tracking**           | Token budget per lane. Warn at 1.5×, pause at 3×. Surfaced in mission reports.                  |
-| **Multi-actor safe**        | Branch-scoped state. Two engineers, one repo. Reset refuses without `--force`.                  |
-
-→ [Every feature, explained with how-to-use + scenarios](docs/concepts/features.md) · [Full pipeline](docs/concepts/workflow.md) · [Lanes](docs/concepts/lanes.md) · [Modes](docs/concepts/modes.md) · [Config](docs/concepts/config.md) · [Audit trail](docs/concepts/audit-trail.md) · [Cost](docs/concepts/cost.md)
+    Review + security files: invitation-accepted-flow-review.md, invitation-accepted-flow-security.md
+    Findings: 3
+    Blocker ledger rows: 1
+
+    ## State
+
+    | Field | Value |
+    |-------|-------|
+    | Wave | 9 |
+    | Tasks | 6/6 done |
+    | Blockers open | 0 |
+    | Heal cycles | 1 |
+    | Tokens used | 14,200 / 20,000 |
+
+## Lanes
+
+Work is sized to the diff — a typo gets no pipeline, an auth migration gets
+all nine waves. Mugiwara itself is free; token usage depends on the lane:
+
+| Lane                | Waves | Typical tokens |
+| ------------------- | :---: | :------------: |
+| Direct (typo)       |   0   |       ~0       |
+| Lean (small bug)    |   2   |      ~4k       |
+| Standard (feature)  |  5–7  |      ~10k      |
+| Full (architecture) | 9–11  |      ~20k      |
+
+Usage tracked in `.mugiwara/state.json` per mission. Budget warns at 1.5×,
+pauses at 3×.
+
+→ [Full cost model](docs/concepts/cost.md)
 
 ## 30-second try
 
@@ -85,10 +107,6 @@ A Standard lane mission (~10k tokens) produces a branch with test-first
 commits, an audit report, a security review, and a ready PR summary — visible
 at every step in your chat.
 
-→ [Full walkthrough](docs/getting-started.md)
-
-## How it works
-
 You ask. The crew routes automatically. **No agent names to memorize, no
 pipeline config to write.**
 
@@ -104,11 +122,41 @@ pipeline config to write.**
 - **Direct agent** when you know exactly what you need — say the name
 - **Slash commands** when you want to drive: `/mugiwara-plan`, `/mugiwara-review`, `/mugiwara-security`, `/mugiwara-ship`, `/mugiwara onboard`
 
-→ [Full workflow walkthrough](docs/concepts/workflow.md)
+→ [Full walkthrough](docs/getting-started.md) · [Full workflow walkthrough](docs/concepts/workflow.md)
+
+## What Mugiwara does
+
+### All features
+
+| Feature                  | What you get                                                                           |
+| ------------------------ | -------------------------------------------------------------------------------------- |
+| **Lane sizing**          | Work auto-sized from `git diff`. Typo = instant fix. Auth migration = full pipeline.   |
+| **Evidence trail**       | `.mugiwara/` workspace: plans, audit reports, quality reports, review findings, blocker ledger. |
+| **Self-healing**         | Brook reads all failures at once, fixes root causes, re-runs verification. ≤3 cycles.  |
+| **Resume from anywhere** | Session lost? Rebuilds from `.mugiwara/state.json`. Continues, never restarts.         |
+| **12 platforms**         | Claude Code, opencode, Copilot, Gemini, Codex, Cursor, Kimi, Pi, Antigravity + CLI.    |
+
+→ All 19 features, with how-to-use + scenarios: [Every feature](docs/concepts/features.md) · [Full pipeline](docs/concepts/workflow.md) · [Lanes](docs/concepts/lanes.md) · [Modes](docs/concepts/modes.md) · [Config](docs/concepts/config.md) · [Audit trail](docs/concepts/audit-trail.md) · [Cost](docs/concepts/cost.md)
+
+## The pipeline
+
+```mermaid
+flowchart TB
+    L0["Luffy<br>Triage"] --> L1["Usopp<br>Brainstorm"] --> L2["Nami<br>Plan"] --> L3["Zoro<br>Execute"] --> L4["Chopper<br>Audit"]
+    L4 --> L5["Sanji<br>Quality"] --> L6["Franky<br>Gates"]
+    L6 --> L7R["Robin<br>Review"]
+    L6 --> L7J["Jinbe<br>Security"]
+    L7R --> L8["Brook<br>Heal"]
+    L7J --> L8
+    L8 --> L9["Luffy<br>Closure"]
+    L8 -. "heal ≤3 cycles" .-> L4
+```
+
+→ [Full pipeline details](docs/concepts/workflow.md)
 
 ## The crew
 
-12 user-facing specialists (+3 internal). Each has role boundaries — auditors
+12 agents (+3 internal). Each has role boundaries — auditors
 and reviewers are read-only. Call them by name or let the pipeline auto-route.
 
 | Agent                | Role                                                                                    |  Permission   |
@@ -136,21 +184,15 @@ and reviewers are read-only. Call them by name or let the pipeline auto-route.
 
 → [Agent details: summoning, boundaries, parameters](docs/concepts/agents.md)
 
-## How much it costs
-
-Mugiwara itself is free. Token usage depends on mission lane:
-
-| Lane                | Waves | Typical tokens |
-| ------------------- | :---: | :------------: |
-| Direct (typo)       |   0   |       ~0       |
-| Lean (small bug)    |   2   |      ~4k       |
-| Standard (feature)  |  5–7  |      ~10k      |
-| Full (architecture) | 9–11  |      ~20k      |
-
-Usage tracked in `.mugiwara/state.json` per mission. Budget warns at 1.5×,
-pauses at 3×.
+## When not to use Mugiwara
 
-→ [Full cost model](docs/concepts/cost.md)
+- **Prototyping or spikes** — use Lane 4, or skip mugiwara entirely.
+- **Unattended multi-hour runs** — the crew runs inline so you can interrupt it.
+  If you want to walk away, superpowers' subagent-driven-development is built
+  for that.
+- **Solo scripts with no review path** — the audit trail has no audience.
+- **Harnesses without agent dispatch** (Gemini, Codex, tier 3) — you get the
+  workflow and the trail, not enforced role boundaries.
 
 ## Configuration
 
@@ -209,6 +251,8 @@ Add to `opencode.json`:
 { "plugin": ["@ionivetech/mugiwara"] }
 ```
 
+Update: `rm -rf ~/.cache/opencode/packages/@ionivetech/mugiwara* && opencode plugin @ionivetech/mugiwara -g` ([details](docs/install/opencode.md#update))
+
 Uninstall: remove `"@ionivetech/mugiwara"` from `opencode.json` plugins array
 
 </details>
@@ -315,20 +359,29 @@ Uninstall: `mugiwara uninstall`
 
 </details>
 
-All platforms get the full crew — 12 agents, 26 skills (+3 internal agents).
-No per-platform feature gaps.
+All platforms get the full crew — 12 agents (+3 internal), 26 skills.
+Enforcement depth varies by harness; see the [harness matrix](docs/reference/harness-matrix.md).
 
 → [Per-platform guides](docs/install/index.md)
 
 ## Update
 
 ```bash
-npm update @ionivetech/mugiwara                           # npm-based
-/plugin update mugiwara                                   # marketplace-based
-# Reinstall with same install command                     # GitHub-based
+# opencode — clear the pinned cache, then reinstall (npm update alone does NOT work)
+rm -rf ~/.cache/opencode/packages/@ionivetech/mugiwara* && opencode plugin @ionivetech/mugiwara -g
+
+# Claude Code — marketplace
+/plugin update mugiwara
+
+# CLI — npm global
+npm i -g @ionivetech/mugiwara@latest
 ```
 
-→ [Full update reference](docs/install/index.md)
+OpenCode pins the resolved version in its own package cache, so `npm update`
+never touches it. Reinstall with the same command for GitHub-based plugins
+(Gemini, Codex, Copilot, Cursor, Kimi, Pi, Antigravity).
+
+→ [Per-platform guides](docs/install/index.md)
 
 ## CLI
 

package/content/skills/mugiwara-brainstorm/SKILL.md

@@ -87,3 +87,10 @@ For UI directions, name slop risks (generic card grids, unmotivated gradients, t
 ## One sharp question rule
 
 If you cannot phrase the question as multiple choice with answerable options, you do not yet understand the decision — read the codebase until you can.
+
+## Red flags
+
+- Collapsing to fewer than three interrogation rounds before handoff.
+- Rubber-stamping ("yes, done") instead of options + trade-offs + recommendation.
+- Guessing a version or library capability without web research.
+- Handing off with a failing validation checklist.

package/content/skills/mugiwara-execution/SKILL.md

@@ -14,8 +14,6 @@ Execute the plan exactly. No silent reordering, no skipping steps, no "close eno
 
 ## Ask before working
 
-By mode (per mode config):
-
 - `guided`: before touching any code, ASK THE USER — auto branch (dedicated mission branch, recommended, keeps `main` clean) or work on the current branch; auto commit per task or commit at user-controlled checkpoints.
 - `semi`/`auto`: auto-create the mission branch per the config `branch` key (default `feature/{type}-{issue}-{slug}`) and auto-commit per task using the config `commit` style (default conventional). No branch/commit ask. Record mode + branch + commit style in the decision log (`.mugiwara/logs/YYYY-MM-DD-<mission>.md`) and in `.mugiwara/results/<mission>/todos.md`.
 
@@ -38,40 +36,57 @@ Before starting: if `.mugiwara/continue.md` exists, resume from its next_action
 1. Read the plan doc fully before touching code.
 2. Build the task graph from `[PARALLEL]`/`[SEQUENTIAL]` markers and depends-on fields.
 3. Contradictory graph (cycle, missing dependency) → escalate to Luffy. Do not guess.
-4. SEQUENTIAL tasks and chains → execute INLINE in the main thread, one at a time, in plan order. The user watches the work happen; no subagent round-trips for ordered work.
+4. SEQUENTIAL tasks and chains → execute INLINE in the main thread, one at a time, in plan order. The user watches the work happen; no subagent round-trips for ordered work — UNLESS context pressure triggers (see Worker dispatch triggers).
 5. Independent `[PARALLEL]` task batches → dispatch WORKER subagents concurrently, one task per worker (host's native task/subagent mechanism). Workers are not crew members. A worker's result returns as a report; summarize inline with evidence pointers before starting the next batch.
 6. Two tasks must never edit the same file concurrently. The plan should prevent this; if it doesn't, serialize them and note the deviation.
 
-## Batch resume
+## Worker dispatch triggers
 
-After each batch, update `.mugiwara/continue.md` next_action to the next task; `[PARALLEL]` batches stay per sub-mission, never crossing a sub-mission boundary.
+1. **Independence** — `[PARALLEL]` batches, concurrent, one task per worker.
+2. **Context pressure** — when `tokens_est` exceeds 60% of `budget`
+   mid-execution, remaining SEQUENTIAL tasks dispatch to workers — one at a
+   time, in plan order. Order is preserved; only the context resets.
 
-## Task batching
+Announce: `⚠ context 62% — remaining tasks run in fresh workers, plan order unchanged.`
+
+The threshold stays relative, never absolute: `tokens_est > 60% × budget`
+(survives model generations), never `tokens_est > 80,000` (obsolete in six
+months). A bigger window raises the threshold; it does not remove it.
+
+## Tier gating & fallback
+
+Real worker dispatch exists only where the harness has subagents — tier 1
+(Claude Code, opencode) plus Copilot. Gate the context-pressure trigger on
+that capability: if the harness cannot dispatch, do not promise fresh workers.
+
+Where workers are unavailable and context pressure crosses the threshold, fall
+back to the mechanism that already exists: write a savepoint, run the
+checkpoint, and suggest a fresh session via `resume`. Announce the fallback so
+the user is not guessing:
 
-Run task work tightly: do the steps without narrating each command or micro-step. Surface ONE per-task result + evidence per task (or per batch) — status, evidence pointer, deviations — in a compact line or table. The checkpoint audits evidence, not commentary; save the blow-by-blow.
+`⚠ context 62% — no worker dispatch on this harness; savepoint written,
+resume in a fresh session (plan order unchanged).`
 
-**Output rule.** Do NOT stream every tool call to the main thread. After each task batch, emit ONLY:
+## Batch resume
+
+After each batch, update `.mugiwara/continue.md` next_action to the next task; `[PARALLEL]` batches stay per sub-mission, never crossing a sub-mission boundary.
 
-```
-T1: ✅ | built + tested | bun run test -- installer
-T2: ✅ | 7 pointers rewritten | grep refs/ → clean
-T3: ✅ | 38/38 tests | bun run test
-```
+## Task batching
 
-Full logs go to `.mugiwara/results/<mission>/01-execution.md`. The main thread shows the summary table only. Tool calls visible below the banner are noise — batch them, squash the output.
+Full protocol: `references/dispatch.md` — output rule, batch report format.
 
 ## Delegation format (parallel workers only)
 
-Sequential work runs inline — no delegation. For every `[PARALLEL]` worker you dispatch, the prompt includes all six fields:
+Full protocol: `references/dispatch.md` — six-field worker prompt. Thin prompts cause thin results.
 
-- TASK — the task body, verbatim from the plan.
-- EXPECTED OUTCOME — what "done" looks like, concrete and checkable.
-- REQUIRED TOOLS — commands and files the subagent will need.
-- MUST DO — the steps in order, including the TDD failing-test-first step.
-- MUST NOT DO — boundaries: files not to touch, configs not to weaken, no silent workarounds.
-- CONTEXT — interfaces consumed/produced, related tasks, mission workspace paths.
+## Surfacing rule
 
-A delegation prompt shorter than ~30 lines is too short — beef it up. Thin prompts cause thin results.
+> **Delegated work is not hidden work.** A worker may run out of view; its
+> result may not. Every worker returns a wave banner, a one-line verdict, and an
+> evidence path into the main thread. The user never clicks into a subagent to
+> know what happened.
+>
+> Isolation is for context and permission, never for autonomy.
 
 ## TDD discipline & user tests
 
@@ -79,19 +94,11 @@ Full protocol: `references/resume-batching.md` — batch-resume, TDD RED-GREEN-R
 
 ## One logical task, one commit
 
-1. Follow the task's steps in order — TDD discipline above: failing test first (watch it fail), implement, watch it pass, refactor while green.
-2. Verify every acceptance criterion; capture command output as evidence.
-3. Commit per LOGICAL task: a task is a meaningful unit of work (a feature, a fix, a refactor) — not a micro-step. Adjacent trivial changes (typo, formatting, a one-line tweak) fold into the neighboring logical task's commit; never one commit per keystroke. If the plan slices tasks finer than a logical change, group adjacent tasks into one commit and note the grouping in the execution report.
-4. Commit only the files that task declared. No task commingles with its neighbors.
-5. Report done (with evidence) or blocked (with reason).
+Commit per LOGICAL task — a feature, fix, or refactor, not a micro-step; verify every acceptance criterion, commit only the task's declared files. Report done (with evidence) or blocked (with reason).
 
 ## Blockers → issues ledger
 
-Blocked → write one row to `.mugiwara/issues/YYYY-MM-DD-<mission>-blockers.md`:
-
-| wave | task | symptom | attempted | help-needed |
-
-Then escalate to Luffy. Never work around a blocker silently.
+Blocked → one row `| wave | task | symptom | attempted | help-needed |` to `.mugiwara/issues/YYYY-MM-DD-<mission>-blockers.md`, then escalate to Luffy. Never work around a blocker silently.
 
 ## Frontend tasks
 
@@ -99,15 +106,7 @@ Any task touching UI markup, styling, or components applies `mugiwara-frontend`
 
 ## Report
 
-After each wave: compact task table (status, evidence pointer, deviations) shown inline in the conversation. Format:
-
-```
-| # | Task | Status | Evidence |
-|---|------|--------|----------|
-| T1 | <title> | ✅/❌ | <command or file> |
-```
-
-Then return to Luffy, who routes to Chopper (Wave 4). Write detailed execution log to `.mugiwara/results/<mission>/01-execution.md`. Never dispatch another crew member.
+After each wave: compact task table (status, evidence pointer, deviations) shown inline in the conversation. Format: `references/dispatch.md` — report table. Then return to Luffy, who routes to Chopper (Wave 4). Write detailed execution log to `.mugiwara/results/<mission>/01-execution.md`. Never dispatch another crew member.
 
 ## Red flags
 
@@ -119,5 +118,6 @@ Then return to Luffy, who routes to Chopper (Wave 4). Write detailed execution l
 - The task's TDD order inverted (implementation before the failing test).
 - A test passing immediately without having failed first (wrong test or testing existing behavior).
 - A commit containing files beyond its declared task, or a wave of micro-commits with no logical grouping.
+- Dispatching a worker whose result is not summarized inline with an evidence path.
 
 All mean: stop, realign to the plan, or escalate to Luffy.

package/content/skills/mugiwara-execution/references/dispatch.md

@@ -0,0 +1,41 @@
+# Dispatch & batching detail
+
+Full detail behind `content/skills/mugiwara-execution/SKILL.md` — the output
+rule, the worker prompt format, and the per-wave report table.
+
+## Task batching
+
+Run task work tightly: do the steps without narrating each command or micro-step. Surface ONE per-task result + evidence per task (or per batch) — status, evidence pointer, deviations — in a compact line or table. The checkpoint audits evidence, not commentary; save the blow-by-blow.
+
+**Output rule.** Do NOT stream every tool call to the main thread. After each task batch, emit ONLY:
+
+```
+T1: ✅ | built + tested | bun run test -- installer
+T2: ✅ | 7 pointers rewritten | grep refs/ → clean
+T3: ✅ | 38/38 tests | bun run test
+```
+
+Full logs go to `.mugiwara/results/<mission>/01-execution.md`. The main thread shows the summary table only. Tool calls visible below the banner are noise — batch them, squash the output.
+
+## Delegation format (parallel workers only)
+
+Sequential work runs inline — no delegation. For every `[PARALLEL]` worker you dispatch, the prompt includes all six fields:
+
+- TASK — the task body, verbatim from the plan.
+- EXPECTED OUTCOME — what "done" looks like, concrete and checkable.
+- REQUIRED TOOLS — commands and files the subagent will need.
+- MUST DO — the steps in order, including the TDD failing-test-first step.
+- MUST NOT DO — boundaries: files not to touch, configs not to weaken, no silent workarounds.
+- CONTEXT — interfaces consumed/produced, related tasks, mission workspace paths.
+
+A delegation prompt shorter than ~30 lines is too short — beef it up. Thin prompts cause thin results.
+
+## Report table
+
+After each wave: compact task table (status, evidence pointer, deviations) shown inline in the conversation. Format:
+
+```
+| # | Task | Status | Evidence |
+|---|------|--------|----------|
+| T1 | <title> | ✅/❌ | <command or file> |
+```