launch-registry
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data, which is a known attack surface, but it includes comprehensive mitigations.
- Ingestion points: The skill processes user context, project files, and potentially untrusted external sources like scraped text, reviews, comments, and emails (SKILL.md).
- Boundary markers: The instructions explicitly command the agent to treat these sources as untrusted input and strictly ignore any instructions embedded within that data (SKILL.md).
- Capability inventory: The skill is capable of recommending external side effects (e.g., publishing or account changes), but it is explicitly forbidden from executing them without user authorization (SKILL.md, references/skill-contract.md).
- Sanitization: The skill requires evidence to be labeled by its source and confidence level (e.g., measured vs. estimated), ensuring that the agent does not treat untrusted inputs as verified facts.
- [SAFE]: The skill demonstrates high security maturity by establishing a "Skill Contract" that mandates read-only defaults, requires authorization for persistent writes, and explicitly prohibits the storage of API keys in source control or skill instructions (references/connectors.md, references/skill-contract.md).
Audit Metadata