project-execution
Pass
Audited by Gen Agent Trust Hub on Jul 26, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill documents a standard software development workflow focused on implementation and verification. It emphasizes following existing project patterns and performing testing at appropriate levels (e.g., UI vs. API).
- [DATA_INGESTION]: The skill is designed to process project-related documents such as requirements, technical designs, and task lists. While this represents an ingestion surface for potentially untrusted data (Indirect Prompt Injection), the instructions include checks for source validity and consistency (SKILL.md, 'Beginning Execution' section), which helps mitigate accidental obedience to embedded instructions.
- [COMMAND_EXECUTION]: The instructions permit running tests, build scripts, linters, and type checkers (SKILL.md, 'Self-Test and Completion' section). These are standard development operations and are constrained to the context of the specific project change being implemented.
- [METADATA]: The YAML frontmatter in both SKILL.md and agents/openai.yaml provides accurate descriptions of the skill's capabilities and does not contain deceptive instructions.
Audit Metadata