Protocol Enforcement (Red-Team Verification)

Priority: P0 (CRITICAL)

Strict guidelines for adversarial verification. Assume the implementation is "guilty" of protocol slippage until proven innocent.

Red-Team Verification Protocol

Before declaring any task "done" or calling notify_user:

1. Adversarial Audit: Search for code patterns that look like "Standard Defaults" (e.g., hardcoded values, generic library calls) where a Project Skill exists.
2. Protocol Check: Ensure the "Pre-Write Audit Log" was present for EVERY write tool call.
3. Execution Bias Check: Ask: "Did I skip a structural constraint to make the code run faster/pass a test?"

The Post-Write Self-Scan

Immediately after a tool call:

• Scan: Read the diff or the file content.
• Match: Check against Anti-Patterns in all active skills.
• Fix: Re-edit immediately if a violation is detected.

Anti-Patterns

• No "Done" Bias: Functional success != Protocol success.
• No Reliance on Memory: Always retrieval-led (Skill view_file) before write.
• No Skipping Protocols: "Small changes" are where most violations happen.

Execution Bias Detection

Look for:

• Local mocks instead of shared fakes.
• Hardcoded styles instead of design tokens.
• Try-catch blocks without standard error handling.
• Missing Pre-Write Audit Log in thoughts.

References

• Violation Examples — concrete red-team scan examples per framework