This directory contains copy-ready preferences.json examples plus a couple of optional security.md overrides.

If you want a complete profile you can copy as a starting point, see ../profiles/deliberate-operator/.

How to use them:

1. Copy one JSON example into your profile as profiles/‹name›/preferences.json.
2. Optionally copy one security example into the same profile as profiles/‹name›/security.md.
3. Adjust role names, prompt directives, and feature flags to match how restrictive or capable you want the profile to be.

Notes:

• security.md cannot be disabled through preferences.json.
• If you create a profile-level security.md, it must contain non-empty content.
• prompts.roles.allow must include orchestrator.
• prompts.roles.deny cannot include orchestrator.

Suggested pairings:

• deliberate-operator.json: A measured, planning-heavy profile that still allows coding and review. Pair well with security-cautious.md.
• review-heavy.json: A lean review and exploration profile that disables several runtime features and restricts coding.
• commercial-operator.json: A high-autonomy, outcome-first profile with broad feature access. Pair well with security-high-autonomy.md.

Security override examples:

• security-cautious.md: Strong guardrails for reversible work, explicit approval boundaries, and careful handling of secrets.
• security-high-autonomy.md: Still safe, but tuned for faster execution when the user clearly wants action.