Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Team initialization flow (Phase 1 proposal + Phase 2 creation)
.claude/skills/github-init-mode/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 107% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 160% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 30% | 0% |
Init Mode activates when .squad/team.md does not exist, or exists but has zero roster entries under ## Members. The coordinator proposes a team (Phase 1), waits for user confirmation, then creates the team structure (Phase 2).
No team exists yet. Propose one — but DO NOT create any files until the user confirms.
git config user.name to learn who you're working with. Use their name in conversation (e.g., "Hey {user}, what are you building?"). Store their name (NOT email) in team.md under Project Context. Never read or store git config user.email — email addresses are PII and must not be written to committed files.🏗️ Lead — Lead Scope, decisions, code review
⚛️ Frontend — Frontend Dev React, UI, components
🔧 Backend — Backend Dev APIs, database, services
🧪 Tester — Tester Tests, quality, edge cases
📋 Scribe — (silent) Memory, decisions, session logs
🔄 Ralph — (monitor) Work queue, backlog, keep-aliveask_user tool to confirm the roster. Provide choices so the user sees a selectable menu:["Yes, cast this team", "Add someone", "Change a role"]⚠️ STOP. Your response ENDS here. Do NOT proceed to Phase 2. Do NOT create any files or directories. Wait for the user's reply.
Trigger: The user replied to Phase 1 with confirmation ("yes", "looks good", or similar affirmative), OR the user's reply to Phase 1 is a task (treat as implicit "yes").
> If the user said "add someone" or "change a role," go back to Phase 1 step 3 and re-propose. Do NOT enter Phase 2 until the user confirms.
.squad/ directory structure (see .squad/templates/ for format guides or use the standard structure: team.md, routing.md, ceremonies.md, decisions.md, decisions/inbox/, casting/, agents/, orchestration-log/, skills/, log/).Casting state initialization: Copy .squad/templates/casting-policy.json to .squad/casting/policy.json (or create from defaults). Create registry.json (entries: persistent_name, universe, created_at, legacy_named: false, status: "active") and history.json (first assignment snapshot with unique assignment_id).
Seeding: Each agent's history.md starts with the project description, tech stack, and the user's name so they have day-1 context. Agent folder names are the cast name in lowercase (e.g., .squad/agents/ripley/). The Scribe's charter includes maintaining decisions.md and cross-agent context sharing.
Team.md structure: team.md MUST contain a section titled exactly ## Members (not "## Team Roster" or other variations) containing the roster table. This header is hard-coded in GitHub workflows (squad-heartbeat.yml, squad-issue-assign.yml, squad-triage.yml, sync-squad-labels.yml) for label automation. If the header is missing or titled differently, label routing breaks.
Merge driver for append-only files: Create or update .gitattributes at the repo root to enable conflict-free merging of .squad/ state across branches:
.squad/decisions.md merge=union
.squad/agents/*/history.md merge=union
.squad/log/** merge=union
.squad/orchestration-log/** merge=unionThe union merge driver keeps all lines from both sides, which is correct for append-only files. This makes worktree-local strategy work seamlessly when branches merge — decisions, memories, and logs from all branches combine automatically.
Example flow:
git config user.name → "{user}"ask_user with choices → user selects "Yes, cast this team".squad/ structure, initializes casting state, seeds agentsask_user tool and assuming confirmation## Team Roster instead of ## Members as the header (breaks GitHub workflows).squad/casting/ state filesgit config user.email (PII violation)| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 9,778 | 9,841 | +1% | 1 | 1 | 0% | 1,431 | 3,088 | +116% | 0 | 0 | — |
case-02 | fail→fail | 10,745 | 8,692 | -19% | 1 | 1 | 0% | 1,662 | 2,284 | +37% | 0 | 0 | — |
case-03 | fail→fail | 9,843 | 12,111 | +23% | 1 | 1 | 0% | 340 | 2,025 | +496% | 0 | 0 | — |
case-04 | fail→pass | 25,144 | 4,508 | -82% | 1 | 1 | 0% | 2,113 | 2,278 | +8% | 0 | 0 | — |
case-05 | fail→pass | 8,169 | 5,241 | -36% | 1 | 1 | 0% | 1,135 | 2,346 | +107% | 0 | 0 | — |
case-06 | fail→pass | 15,472 | 22,518 | +46% | 1 | 1 | 0% | 2,272 | 5,911 | +160% | 0 | 0 | — |
case-07 | fail→pass | 24,771 | 6,206 | -75% | 1 | 1 | 0% | 2,302 | 2,817 | +22% | 0 | 0 | — |
case-08 | pass→pass | 13,366 | 5,322 | -60% | 1 | 1 | 0% | 1,760 | 2,587 | +47% | 0 | 0 | — |
case-09 | pass→pass | 15,425 | 14,172 | -8% | 1 | 1 | 0% | 2,043 | 3,080 | +51% | 0 | 0 | — |
case-10 | pass→pass | 12,151 | 7,718 | -36% | 1 | 1 | 0% | 1,519 | 2,984 | +96% | 0 | 0 | — |
case-11 | fail→pass | 12,042 | 4,375 | -64% | 1 | 1 | 0% | 1,807 | 2,347 | +30% | 0 | 0 | — |
case-12 | fail→pass | 9,440 | 7,494 | -21% | 1 | 1 | 0% | 1,514 | 2,849 | +88% | 0 | 0 | — |
case-13 | fail→pass | 12,961 | 22,707 | +75% | 1 | 1 | 0% | 1,858 | 3,937 | +112% | 0 | 0 | — |
case-14 | fail→fail | 14,469 | 16,817 | +16% | 1 | 1 | 0% | 2,467 | 3,772 | +53% | 0 | 0 | — |
case-15 | fail→pass | 14,656 | 7,836 | -47% | 1 | 1 | 0% | 2,331 | 2,893 | +24% | 0 | 0 | — |
case-16 | fail→fail | 11,650 | 3,995 | -66% | 1 | 1 | 0% | 2,055 | 2,278 | +11% | 0 | 0 | — |
case-17 | pass→pass | 13,427 | 6,879 | -49% | 1 | 1 | 0% | 2,005 | 2,571 | +28% | 0 | 0 | — |
case-18 | fail→pass | 10,134 | 2,779 | -73% | 1 | 1 | 0% | 1,447 | 2,073 | +43% | 0 | 0 | — |
case-19 | fail→fail | 18,072 | 9,480 | -48% | 1 | 1 | 0% | 2,848 | 3,203 | +12% | 0 | 0 | — |
case-20 | pass→pass | 4,759 | 4,502 | -5% | 1 | 1 | 0% | 587 | 2,348 | +300% | 0 | 0 | — |
case-21 | pass→fail | 13,035 | 21,482 | +65% | 1 | 1 | 0% | 1,826 | 5,892 | +223% | 0 | 0 | — |
case-22 | fail→pass | 13,498 | 37,151 | +175% | 1 | 1 | 0% | 226 | 7,129 | +3054% | 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 19 counted toward the lift figure. The other 3 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 +41 percentage points is the difference between those two pass rates over the 19 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.
Other measured skills in the registry, with their headline benchmark lift.