Description
Extend agent instructions to explicitly define:
- What agents must NOT do (e.g., add new APIs, bypass proxy, hardcode secrets)
- How agents should handle environment variables
- Rules for network calls (only via proxy)
Acceptance Criteria
- Explicit guardrails documented
- Security-sensitive areas clearly marked
- Instructions are unambiguous for autonomous agents
Description
Extend agent instructions to explicitly define:
Acceptance Criteria