Agent-Skills.md

Agent Skills: Protocol Enforcement (Red-Team Verification)

Standards for Red-Team verification and adversarial protocol audit. Use when verifying tasks, performing self-scans, or checking for protocol violations. Load as composite for all sessions. (triggers: verify done, protocol check, self-scan, pre-write audit, task complete, audit violations, retrospective, scan, red-team)

UncategorizedID: hoangnguyen0403/agent-skills-standard/common-protocol-enforcement

Repository

hoangnguyen0403/agent-skills-standard
HoangNguyen0403License: Apache-2.0
362103

Install this agent skill to your local

pnpm dlx add-skill https://github.com/HoangNguyen0403/agent-skills-standard/tree/HEAD/.agent/skills/common/common-protocol-enforcement

Skill Files

Browse the full folder contents for common-protocol-enforcement.

Download Skill

Loading file tree…

.agent/skills/common/common-protocol-enforcement/SKILL.md

Skill Metadata

Name
common-protocol-enforcement
Description
"Standards for Red-Team verification and adversarial protocol audit. Use when verifying tasks, performing self-scans, or checking for protocol violations. Load as composite for all sessions. (triggers: verify done, protocol check, self-scan, pre-write audit, task complete, audit violations, retrospective, scan, red-team)"

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