Claude-Code-Workflow/.claude/skills/team-review/SKILL.md

name, description, allowed-tools

name	description	allowed-tools
team-review	Unified team skill for code review. Uses team-worker agent architecture with role-spec files. 3-role pipeline: scanner, reviewer, fixer. Triggers on team-review.	TeamCreate(*), TeamDelete(*), SendMessage(*), TaskCreate(*), TaskUpdate(*), TaskList(*), TaskGet(*), Task(*), AskUserQuestion(*), Read(*), Write(*), Edit(*), Bash(*), Glob(*), Grep(*), mcp__ace-tool__search_context(*)

Team Review

Unified team skill: code scanning, vulnerability review, optimization suggestions, and automated fix. All team members invoke with --role=xxx to route to role-specific execution.

Architecture

+---------------------------------------------------+
|  Skill(skill="team-review")                        |
|  args="<task-description>"                         |
+-------------------+-------------------------------+
                    |
         Orchestration Mode (auto -> coordinator)
                    |
              Coordinator (inline)
              Phase 0-5 orchestration
                    |
    +-------+-------+-------+
    v       v       v
 [tw]    [tw]    [tw]
scann-  review- fixer
er      er

(tw) = team-worker agent

Role Router

Input Parsing

Parse $ARGUMENTS to extract --role. If absent → Orchestration Mode (auto route to coordinator).

Role Registry

Role	Spec	Task Prefix	Inner Loop
coordinator	roles/coordinator/role.md	(none)	-
scanner	role-specs/scanner.md	SCAN-*	false
reviewer	role-specs/reviewer.md	REV-*	false
fixer	role-specs/fixer.md	FIX-*	true

⚠️ COMPACT PROTECTION: 角色文件是执行文档，不是参考资料。当 context compression 发生后，角色指令仅剩摘要时，必须立即 Read 对应 role.md 重新加载后再继续执行。不得基于摘要执行任何 Phase。

Dispatch

1. Extract --role from arguments
2. If no --role → route to coordinator (Orchestration Mode)
3. Look up role in registry → Read the role file → Execute its phases

Orchestration Mode

When invoked without --role, coordinator auto-starts. User just provides target description.

Invocation: Skill(skill="team-review", args="<target-path>")

Lifecycle:

User provides scan target
  → coordinator Phase 1-3: Parse flags → TeamCreate → Create task chain
  → coordinator Phase 4: spawn first batch workers (background) → STOP
  → Worker executes → SendMessage callback → coordinator advances next step
  → Loop until pipeline complete → Phase 5 report

User Commands (wake paused coordinator):

Command	Action
check / status	Output execution status graph, no advancement
resume / continue	Check worker states, advance next step

---

Pipeline (CP-1 Linear)

coordinator dispatch
  → SCAN-* (scanner: toolchain + LLM scan)
  → REV-*  (reviewer: deep analysis + report)
  → [user confirm]
  → FIX-*  (fixer: plan + execute + verify)

Cadence Control

Beat model: Event-driven, each beat = coordinator wake → process → spawn → STOP.

Beat Cycle (single beat)
═══════════════════════════════════════════════════════════
  Event                   Coordinator              Workers
───────────────────────────────────────────────────────────
  callback/resume ──→ ┌─ handleCallback ─┐
                      │  mark completed   │
                      │  check pipeline   │
                      ├─ handleSpawnNext ─┤
                      │  find ready tasks │
                      │  spawn workers ───┼──→ [Worker A] Phase 1-5
                      └─ STOP (idle) ─────┘         │
                                                     │
  callback ←─────────────────────────────────────────┘
  (next beat)              SendMessage + TaskUpdate(completed)
═══════════════════════════════════════════════════════════

Pipeline beat view:

Review Pipeline (3 beats, linear with user checkpoint)
──────────────────────────────────────────────────────────
Beat  1         2         ⏸         3
      │         │         │         │
      SCAN → REV ──→ [confirm] → FIX
      ▲                              ▲
   pipeline                       pipeline
    start                          done

SCAN=scanner  REV=reviewer  FIX=fixer

Checkpoints:

Trigger	Location	Behavior
Review→Fix transition	REV-* complete	Pause, present review report, wait for user resume to confirm fix
Quick mode (-q)	After SCAN-*	Pipeline ends after scan, no review/fix
Fix-only mode (--fix)	Entry	Skip scan/review, go directly to fixer

Stall Detection (coordinator handleCheck executes):

Check	Condition	Resolution
Worker no response	in_progress task no callback	Report waiting task list, suggest user resume
Pipeline deadlock	no ready + no running + has pending	Check blockedBy dependency chain, report blocking point

Task Metadata Registry

Task ID	Role	Phase	Dependencies	Description
SCAN-001	scanner	scan	(none)	Toolchain + LLM code scanning
REV-001	reviewer	review	SCAN-001	Deep analysis and review report
FIX-001	fixer	fix	REV-001 + user confirm	Plan + execute + verify fixes

---

Shared Infrastructure

Worker Phase 1: Task Discovery (shared by all workers)

Every worker executes the same task discovery flow on startup:

1. Call TaskList() to get all tasks
2. Filter: subject matches this role's prefix + owner is this role + status is pending + blockedBy is empty
3. No tasks → idle wait
4. Has tasks → TaskGet for details → TaskUpdate mark in_progress

Resume Artifact Check (prevent duplicate output after resume):

• Check whether this task's output artifact already exists
• Artifact complete → skip to Phase 5 report completion
• Artifact incomplete or missing → normal Phase 2-4 execution

Worker Phase 5: Report (shared by all workers)

Standard reporting flow after task completion:

1. Message Bus: Call mcp__ccw-tools__team_msg to log message
   • Parameters: operation="log", session_id=, from=, type=, data={ref: ""}
   • to and summary auto-defaulted -- do NOT specify explicitly
   • CLI fallback: ccw team log --session-id <session-id> --from <role> --type <type> --json
2. SendMessage: Send result to coordinator
3. TaskUpdate: Mark task completed
4. Loop: Return to Phase 1 to check next task

Wisdom Accumulation (all roles)

Cross-task knowledge accumulation. Coordinator creates wisdom/ directory at session initialization.

Directory:

<session-folder>/wisdom/
├── learnings.md      # Patterns and insights
├── decisions.md      # Architecture and design decisions
├── conventions.md    # Codebase conventions
└── issues.md         # Known risks and issues

Worker Load (Phase 2): Extract Session: <path> from task description, read wisdom directory files. Worker Contribute (Phase 4/5): Write this task's discoveries to corresponding wisdom files.

Role Isolation Rules

Allowed	Forbidden
Process tasks with own prefix	Process tasks with other role prefixes
SendMessage to coordinator	Communicate directly with other workers
Use tools declared in Toolbox	Create tasks for other roles
Delegate to commands/ files	Modify resources outside own responsibility

Coordinator additional restrictions: Do not write/modify code directly, do not call implementation subagents, do not execute analysis/test/review directly.

Team Configuration

Setting	Value
Team name	review
Session directory	.workflow/.team/RV-<slug>-<date>/
Shared memory	.msg/meta.json in session dir
Team config	specs/team-config.json
Finding schema	specs/finding-schema.json
Dimensions	specs/dimensions.md

---

Coordinator Spawn Template

v5 Worker Spawn (all roles)

When coordinator spawns workers, use team-worker agent with role-spec path:

Task({
  subagent_type: "team-worker",
  description: "Spawn <role> worker",
  team_name: "review",
  name: "<role>",
  run_in_background: true,
  prompt: `## Role Assignment
role: <role>
role_spec: .claude/skills/team-review/role-specs/<role>.md
session: <session-folder>
session_id: <session-id>
team_name: review
requirement: <task-description>
inner_loop: <true|false>

Read role_spec file to load Phase 2-4 domain instructions.
Execute built-in Phase 1 (task discovery) -> role-spec Phase 2-4 -> built-in Phase 5 (report).`
})

Inner Loop roles (fixer): Set inner_loop: true.

Single-task roles (scanner, reviewer): Set inner_loop: false.

Usage

# Via coordinator (auto pipeline)
Skill(skill="team-review", args="src/auth/**")                    # scan + review
Skill(skill="team-review", args="--full src/auth/**")             # scan + review + fix
Skill(skill="team-review", args="--fix .review/review-*.json")    # fix only
Skill(skill="team-review", args="-q src/auth/**")                 # quick scan only

# Direct role invocation
Skill(skill="team-review", args="--role=scanner src/auth/**")
Skill(skill="team-review", args="--role=reviewer --input scan-result.json")
Skill(skill="team-review", args="--role=fixer --input fix-manifest.json")

# Flags (all modes)
--dimensions=sec,cor,perf,maint    # custom dimensions (default: all 4)
-y / --yes                         # skip confirmations
-q / --quick                       # quick scan mode
--full                             # full pipeline (scan → review → fix)
--fix                              # fix mode only

---

Completion Action

When the pipeline completes (all tasks done, coordinator Phase 5):

AskUserQuestion({
  questions: [{
    question: "Review pipeline complete. What would you like to do?",
    header: "Completion",
    multiSelect: false,
    options: [
      { label: "Archive & Clean (Recommended)", description: "Archive session, clean up tasks and team resources" },
      { label: "Keep Active", description: "Keep session active for follow-up work or inspection" },
      { label: "Export Results", description: "Export deliverables to a specified location, then clean" }
    ]
  }]
})

Choice	Action
Archive & Clean	Update session status="completed" -> TeamDelete(review) -> output final summary
Keep Active	Update session status="paused" -> output resume instructions: Skill(skill="team-review", args="resume")
Export Results	AskUserQuestion for target path -> copy deliverables -> Archive & Clean

---

Session Directory

.workflow/.team/RV-<slug>-<YYYY-MM-DD>/
├── .msg/
│   ├── messages.jsonl          # Message bus log
│   └── meta.json               # Session state + cross-role state
├── wisdom/                     # Cross-task knowledge
│   ├── learnings.md
│   ├── decisions.md
│   ├── conventions.md
│   └── issues.md
├── scan/                       # Scanner output
│   └── scan-results.json
├── review/                     # Reviewer output
│   └── review-report.json
└── fix/                        # Fixer output
    └── fix-manifest.json

Error Handling

Scenario	Resolution
Unknown --role value	Error with available role list
Missing --role arg	Orchestration Mode → auto route to coordinator
Role file not found	Error with expected file path (roles//role.md)
Invalid flags	Warn and continue with defaults
No target specified (no --role)	AskUserQuestion to clarify

Execution Rules

1. Parse first: Extract --role and flags from $ARGUMENTS before anything else
2. Progressive loading: Read ONLY the matched role.md, not all four
3. Full delegation: Role.md owns entire execution -- do not add logic here
4. Self-contained: Each role.md includes its own message bus, task lifecycle, toolbox
5. DO NOT STOP: Continuous execution until role completes all 5 phases