Make a coding agent work safely in a repo

# Make a coding agent work safely in a repo

You are an engineering lead setting rules for an AI coding agent. I will give you a repo, task, risk profile, and workflow. Create safe operating instructions that keep the agent useful without letting it wander.

## Inputs

Repo context: {{repo_context||Framework, folders, tests, package manager, deployment, conventions, or repo tree.}}
Task type: {{task_type||Bug fix, feature, refactor, audit, test writing, docs, design implementation, migration, etc.}}
Risk profile: {{risk_profile||Production app, payments, auth, data, user content, low risk, unknown, regulatory, security, etc.}}
Human preferences: {{human_preferences||How much autonomy, when to ask, what tools to use, files to avoid, commit style, or update cadence.}}

## Output

**1. Agent mission.** One clear sentence.

**2. Allowed scope.** Files, modules, commands, and behaviors the agent may touch.

**3. Forbidden scope.** Destructive commands, secrets, migrations, unrelated refactors, formatting churn, dependencies, or production changes.

**4. Work loop.** Inspect, plan, edit, verify, report. Include when to stop and ask.

**5. Verification commands.** Tests, lint, build, screenshots, manual checks, or safe fallbacks.

**6. Review checklist.** What the human should check before merging.

## Rules

[Preview stops here. Unlock the Pro library for the full rules, guardrails, examples, and copyable file.]