Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Unified team skill for architecture optimization. Uses team-worker agent architecture with role directories for domain logic. Coordinator orchestrates pipeline, workers are team-worker agents. Triggers on "team arch-opt".
.claude/skills/catlog22-team-arch-opt/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 72% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 119% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -6% | 0% |
> Agent timeout: spawn_agent 异步执行且无内置超时 — 除明确短任务外一律 spawn_agent 后立即 wait_agent({ timeout_ms: 3600000 })(上限 1 小时)阻塞等待,绝不依赖 30000 默认值;timed_out: true 且 Agent 未完成时再次 wait_agent 续等,不丢弃。批量场景使用 spawn_agents_on_csv({ max_runtime_seconds: 3600, ... })。
<required_reading> @~/.maestro/workflows/run-mode-lite.md </required_reading>
Orchestrate multi-agent architecture optimization: analyze codebase → design refactoring plan → implement changes → validate improvements → review code quality.
spawn_agent({ task_name: "team_arch_opt", message: "Execute skill team-arch-opt, args: task description" })
|
SKILL.md (this file) = Router
|
+--------------+--------------+
| |
no --role flag --role <name>
| |
Coordinator Worker
roles/coordinator/role.md roles/<name>/role.md
|
+-- analyze → dispatch → spawn workers → STOP
|
+-------+-------+-------+-------+
v v v v v
[analyzer][designer][refactorer][validator][reviewer]| Role | Path | Prefix | Inner Loop | |------|------|--------|------------| | coordinator | roles/coordinator/role.md | — | — | | analyzer | roles/analyzer/role.md | ANALYZE- | false | | designer | [roles/designer/role.md](roles/designer/role.md) | DESIGN- | false | | refactorer | roles/refactorer/role.md | REFACTOR-, FIX- | true | | validator | roles/validator/role.md | VALIDATE- | false | | reviewer | [roles/reviewer/role.md](roles/reviewer/role.md) | REVIEW-, QUALITY- | false |
.workflow/codebase/ARCHITECTURE.md exists, read for module boundariesmaestro load --type spec --category arch — load arch constraints as shared contextmaestro load --type spec --category coding — load coding constraints as shared contextmaestro search "architecture optimization refactor" --json — top 5 entries as prior contextParse $ARGUMENTS:
--role <name> → Read roles/<name>/role.md, execute Phase 2-4--role → @roles/coordinator/role.md, execute entry routerTAO{run_dir}/work/team/maestro delegate --mode analysis (read-only), maestro delegate --mode write (modifications)mcp__maestro__team_msg(session_id=<run-id>, ...)Coordinator spawns workers using this template:
spawn_agent({
subagent_type: "team-worker",
description: "Spawn <role> worker",
team_name: "arch-opt",
name: "<role>",
run_in_background: true,
prompt: `## Role Assignment
role: <role>
role_spec: <skill_root>/roles/<role>/role.md
session: {run_dir}/work/team
session_id: <run-id>
team_name: arch-opt
requirement: <task-description>
inner_loop: <true|false>
## Progress Milestones
session_id: <run-id>
Report progress via team_msg at natural phase boundaries (context loaded -> core work done -> verification).
Report blockers immediately via team_msg type="blocker".
Report completion via team_msg type="task_complete" after final send_message.
Read role_spec file (@<skill_root>/roles/<role>/role.md) to load Phase 2-4 domain instructions.
Execute built-in Phase 1 (task discovery) -> role Phase 2-4 -> built-in Phase 5 (report).`
})Inner Loop roles (refactorer): Set inner_loop dynamically — true for single mode, false for fan-out/independent (parallel branches). Single-task roles (analyzer, designer, validator, reviewer): Set inner_loop: false.
| Command | Action | |---------|--------| | check / status | View execution status graph (branch-grouped), no advancement | | resume / continue | Check worker states, advance next step | | revise <TASK-ID> [feedback] | Revise specific task + cascade downstream | | feedback <text> | Analyze feedback impact, create targeted revision chain | | recheck | Re-run quality check | | improve [dimension] | Auto-improve weakest dimension |
{run_dir}/work/team/
├── team-session.json # Session metadata + status + parallel_mode
├── task-analysis.json # Coordinator analyze output
├── {run_dir}/outputs/ # Run deliverables (via maestro run)
│ ├── architecture-baseline.json # Analyzer: pre-refactoring metrics
│ ├── architecture-report.md # Analyzer: ranked structural issue findings
│ ├── refactoring-plan.md # Designer: prioritized refactoring plan
│ ├── validation-results.json # Validator: post-refactoring validation
│ ├── review-report.md # Reviewer: code review findings
│ ├── aggregate-results.json # Fan-out/independent: aggregated results
│ ├── branches/ # Fan-out mode branch artifacts
│ │ └── B{NN}/
│ │ ├── refactoring-detail.md
│ │ ├── validation-results.json
│ │ └── review-report.md
│ └── pipelines/ # Independent mode pipeline artifacts
│ └── {P}/
│ └── ...
├── explorations/
│ ├── cache-index.json # Shared explore cache
│ └── <hash>.md
├── wisdom/
│ └── patterns.md # Discovered patterns and conventions
├── {run_dir}/evidence/discussions/
│ ├── DISCUSS-REFACTOR.md
│ └── DISCUSS-REVIEW.md
└── .msg/
├── messages.jsonl # Message bus log
└── meta.json # Session state + cross-role state| Scenario | Resolution | |----------|------------| | Unknown command | Error with available command list | | Role not found | Error with role registry | | CLI tool fails | Worker fallback to direct implementation | | Fast-advance conflict | Coordinator reconciles on next callback | | Completion action fails | Default to Keep Active | | consensus_blocked HIGH | Coordinator creates revision task or pauses pipeline | | Branch fix cycle >= 3 | Escalate only that branch to user, others continue | | max_branches exceeded | Coordinator truncates to top N at CP-2.5 |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 2,424 | 6,475 | +167% | 1 | 1 | 0% | 312 | 2,370 | +660% | 0 | 0 | — |
case-02 | fail→fail | 16,990 | 6,009 | -65% | 1 | 1 | 0% | 2,793 | 2,059 | -26% | 0 | 0 | — |
case-03 | fail→pass | 12,811 | 3,400 | -73% | 1 | 1 | 0% | 2,025 | 2,475 | +22% | 0 | 0 | — |
case-04 | fail→pass | 7,898 | 1,723 | -78% | 1 | 1 | 0% | 1,242 | 2,133 | +72% | 0 | 0 | — |
case-05 | fail→pass | 13,668 | 8,786 | -36% | 1 | 1 | 0% | 2,219 | 3,401 | +53% | 0 | 0 | — |
case-06 | fail→pass | 7,761 | 4,342 | -44% | 1 | 1 | 0% | 1,191 | 2,610 | +119% | 0 | 0 | — |
case-07 | pass→pass | 10,620 | 1,980 | -81% | 1 | 1 | 0% | 1,723 | 2,164 | +26% | 0 | 0 | — |
case-08 | fail→pass | 16,084 | 3,139 | -80% | 1 | 1 | 0% | 2,545 | 2,382 | -6% | 0 | 0 | — |
case-09 | fail→pass | 5,995 | 2,459 | -59% | 1 | 1 | 0% | 918 | 2,206 | +140% | 0 | 0 | — |
case-10 | pass→pass | 8,636 | 6,961 | -19% | 1 | 1 | 0% | 1,200 | 2,763 | +130% | 0 | 0 | — |
case-11 | fail→pass | 11,187 | 2,811 | -75% | 1 | 1 | 0% | 1,979 | 2,355 | +19% | 0 | 0 | — |
case-12 | pass→pass | 9,590 | 2,550 | -73% | 1 | 1 | 0% | 1,504 | 2,261 | +50% | 0 | 0 | — |
case-13 | pass→pass | 6,768 | 1,596 | -76% | 1 | 1 | 0% | 1,061 | 1,997 | +88% | 0 | 0 | — |
case-14 | fail→pass | 8,573 | 1,645 | -81% | 1 | 1 | 0% | 1,359 | 2,113 | +55% | 0 | 0 | — |
case-15 | pass→pass | 4,626 | 3,808 | -18% | 1 | 1 | 0% | 741 | 2,430 | +228% | 0 | 0 | — |
case-16 | fail→pass | 11,036 | 3,167 | -71% | 1 | 1 | 0% | 1,697 | 2,345 | +38% | 0 | 0 | — |
case-21 | fail→fail | 20,384 | 5,039 | -75% | 1 | 1 | 0% | 5,036 | 2,104 | -58% | 0 | 0 | — |
case-17 | fail→pass | 4,755 | 2,405 | -49% | 1 | 1 | 0% | 781 | 2,144 | +175% | 0 | 0 | — |
case-18 | fail→pass | 5,048 | 2,599 | -49% | 1 | 1 | 0% | 847 | 2,235 | +164% | 0 | 0 | — |
case-19 | pass→pass | 4,364 | 2,098 | -52% | 1 | 1 | 0% | 697 | 2,185 | +213% | 0 | 0 | — |
case-20 | fail→fail | 21,436 | 6,358 | -70% | 1 | 1 | 0% | 4,969 | 2,200 | -56% | 0 | 0 | — |
case-22 | fail→fail | 15,206 | 9,303 | -39% | 1 | 1 | 0% | 2,925 | 2,267 | -22% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 17 counted toward the lift figure. The other 5 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +50 percentage points is the difference between those two pass rates over the 17 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/3/2026 | +50% |
Other measured skills in the registry, with their headline benchmark lift.