Team Constitution

team-constitution

What this skill does

Interactively create or amend the team constitution at {TEAM_AI_DIRECTIVES}/context_modules/constitution.md.

The scaffold written by team-setup (Mode 3) contains only a placeholder ("No team-wide principles defined yet"). This skill turns that placeholder into a real constitution through guided elicitation:

• Propose principles informed by team context (existing rules, CDRs, AGENTS.md)
• Review each principle interactively — confirm, edit, or reject
• Write the constitution in the established team format (OKF v0.2 frontmatter + numbered principles + Governance section)

What this skill is NOT:

• Not a port of spec-kit's spec.constitution template machinery — no placeholder tokens, no hooks, no extensions.yml, no template propagation, no Sync Impact Report
• Not semantic versioning — the team constitution is versioned by git history in the team AI directives repo; freshness is tracked by OKF v0.2 frontmatter (generated, verified, stale_after) via /team-repair
• Not for project-level constitutions (docs/adlc/memory/constitution.md, legacy .adlc/memory/constitution.md fallback) — those remain spec.constitution territory when spec-kit is in play

When to use

• After team-setup Mode 3: the scaffold wrote a placeholder constitution — fill it with real principles
• New team AI directives (cloned or existing): constitution is missing or still a placeholder
• Amending principles: the team wants to add, reword, or remove principles

When NOT to use

• Project-level constitution: use spec-kit's spec.constitution for docs/adlc/memory/constitution.md (legacy .adlc/memory/constitution.md fallback)
• Constitution changes via CDR flow: if a constitution change was proposed as a CDR, let /team-levelup handle it
• team AI directives not configured: run /team-setup first

Process

User Input

$ARGUMENTS

You MUST consider the user input before proceeding (if not empty).

Examples of User Input:

• "Focus on testing and security principles" — steer proposals toward those domains
• "Amend principle 3" — jump straight to amending a specific principle
• "Add a principle about AI code review" — append a specific principle
• Empty input: full elicitation flow (create or amend, auto-detected)

Role & Context

You are acting as a Team Governance Facilitator — helping the team articulate its non-negotiable principles. Your role involves:

• Proposing principles grounded in the team's actual directives and practices (not generic boilerplate)
• Ensuring every principle is declarative, testable, and has a clear rationale
• Keeping the constitution concise — principles are rules, not essays

Outline

1. Environment Setup (Phase 0): Resolve paths, detect constitution state
2. Load Team Context (Phase 1): Read existing constitution + team directives to inform proposals
3. Elicit Principles (Phase 2): Interactive create/amend loop
4. Governance Section (Phase 3): Compose the lightweight governance paragraph
5. Write Constitution (Phase 4): Write the file with OKF v0.2 frontmatter
6. Commit & Summary (Phase 5): Offer commit, report results

Execution Steps

Phase 0: Environment Setup

Run the setup script from the skill's base directory:

scripts/bash/setup-team-constitution.sh

Parse JSON for REPO_ROOT, TEAM_AI_DIRECTIVES, CONSTITUTION_FILE, CONSTITUTION_STATE, TD_IS_GIT, TD_CLEAN.

If the setup script is unavailable or fails, resolve manually:

1. REPO_ROOT — walk up from cwd to find .adlc/, or git rev-parse --show-toplevel, or pwd.
2. TEAM_AI_DIRECTIVES — TEAM_AI_DIRECTIVES env var, then .adlc/init-options.json → team_ai_directives, then REPO_ROOT/team-ai-directives.
3. CONSTITUTION_FILE — {TEAM_AI_DIRECTIVES}/context_modules/constitution.md
4. CONSTITUTION_STATE:
   • missing — file does not exist
   • placeholder — file exists and contains "No team-wide principles defined yet"
   • populated — file exists with real content
5. TD_IS_GIT — git -C "$TEAM_AI_DIRECTIVES" rev-parse --is-inside-work-tree (exit 0 = true)
6. TD_CLEAN — git -C "$TEAM_AI_DIRECTIVES" status --porcelain (empty = clean)

If TEAM_AI_DIRECTIVES is not configured:

Team AI directives repository not configured.
Run: team-setup
Or set: export TEAM_AI_DIRECTIVES=/path/to/team-ai-directives

Phase 1: Load Team Context

Read the following to inform principle proposals:

1. Existing constitution at {CONSTITUTION_FILE} (if populated — you will amend; if placeholder/missing — you will create)
2. Team rules index: {TEAM_AI_DIRECTIVES}/CDR.md — accepted rule/persona/example descriptors reveal what the team already cares about
3. Existing rules: skim {TEAM_AI_DIRECTIVES}/context_modules/rules/ file names and headings
4. AGENTS.md at {TEAM_AI_DIRECTIVES}/AGENTS.md — loading order and skill usage expectations

Principles must be grounded in what you find — e.g., if the rules directory is heavy on security modules, a "Security by Default" principle is evidence-based, not boilerplate. If the team AI directives is nearly empty, fall back to a small starter set and say so explicitly.

Phase 2: Elicit Principles

Create mode (missing or placeholder)

1. Propose a starter set (3–7 principles) based on Phase 1 findings. For each principle, present:

### Proposed Principle N: {Name}

**Statement**: {Declarative, testable rule — one or two sentences}

**Rationale**: {Why this principle exists; what failure it prevents}

**Grounding**: {What team context motivated this — or "starter suggestion, no team AI directives evidence yet"}

Reply: [Y] accept / [e] edit / [n] reject

2. For each response:

   • Y: add to the accepted list
   • e: apply the user's edits, present the revised principle for confirmation
   • n: drop it

3. After the starter set, ask: "Any additional principles?" Loop until the user is done.