| name | agent-guidance-boundary-guard |
| description | Use when auditing SHAFT agent guidance, host adapters, skill bridges, guidance budgets, Codex config, or Memory setup. |
Agent Guidance Boundary Guard
Use for agent guidance and memory/config surfaces. Keep AGENTS.md canonical and host adapters thin.
Workflow
- For new or renamed skills, update the bridge in
.agents/skills/**, the canonical .github/skills/** playbook, .github/skills/README.md, and scripts/ci/agent_guidance_budget.json.
- Keep
.agents skill bodies under their configured budget and include agents/openai.yaml with allow_implicit_invocation: true.
- For
.codex or .memory changes, preserve the restricted Memory MCP setup and avoid routine diary/end-session saves.
- Run
python3 scripts/ci/validate_agent_setup.py --skip-external; run the full command when network/npm access is stable.
- If refresh automation changes guidance, check
.github/workflows/refresh-agent-instructions.yml allowlist before widening scope.
Output
List guidance surfaces changed, budget/metadata changes, validation command results, and any intentional budget headroom.
