project-skill-protocol

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Node.js script at .claude/skills/sync-codex/scripts/run-codex-sync.mjs to synchronize environment mirrors after updates. This is an internal framework operation.
  • [PROMPT_INJECTION]: Manages instruction overlays that are injected into other skills, acting as a potential surface for indirect prompt injection. 1. Ingestion points: Reads project-specific rule definitions from docs/project-protocols/ and the index file. 2. Boundary markers: Employs a mandatory PROPOSAL GATE where users must approve drafted rules before they are saved. 3. Capability inventory: Capable of writing to project documentation and triggering local synchronization scripts. 4. Sanitization: Enforces an ADDITIVE ONLY policy that refuses rules attempting to bypass framework safety gates, git discipline, or user-confirmation requirements.
  • [DATA_EXPOSURE]: Interacts with project documentation and CLAUDE.md. The registry contract explicitly prohibits storing credentials or secrets within overlay bodies.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 08:23 PM
Security Audit — agent-trust-hub — project-skill-protocol