Quickstart

Start with the repo-level guide and PR template.

curl -fsSL https://raw.githubusercontent.com/rogerchappel/agentic-team-playbook/main/templates/AGENTS.md -o AGENTS.md
mkdir -p .github
curl -fsSL https://raw.githubusercontent.com/rogerchappel/agentic-team-playbook/main/.github/pull_request_template.md -o .github/pull_request_template.md

Then edit AGENTS.md:

• replace the repo purpose
• choose the repo type
• list real commands
• list risk areas
• add repo-specific verification

Global Agent Policy

For Codex-style global instructions:

mkdir -p ~/.codex
curl -fsSL https://raw.githubusercontent.com/rogerchappel/agentic-team-playbook/main/docs/global-agent-operating-policy.md -o ~/.codex/AGENTS.md

Runtime Adapters

Keep AGENTS.md as the shared source of truth, then add a small adapter file when a runtime expects one.

Claude Code:

cat > CLAUDE.md <<'EOF'
# Claude Code Instructions

Follow AGENTS.md first. It defines the branch-first workflow, atomic commit policy, risk escalation rules, verification requirement, and review pack format for this repository.
EOF

Gemini CLI:

cat > GEMINI.md <<'EOF'
# Gemini CLI Instructions

Follow AGENTS.md first. It defines the branch-first workflow, atomic commit policy, risk escalation rules, verification requirement, and review pack format for this repository.
EOF

OpenCode can use the root AGENTS.md directly.

What Good Looks Like

Agents should start tasks by stating:

1. objective
2. expected blast radius
3. files likely to change
4. commit plan
5. verification plan
6. risk level

They should end with a review pack.

Minimal Review Pack

## Review Pack
Repo:
Branch:
PR:
Task:
Status:
Summary:
Commits:
Files changed:
Verification:
Risk level:
Rollback plan:
Human decision needed:
Next recommended task: