skill-registry
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted metadata from project files which can lead to indirect prompt injection.
- Ingestion points: Steps 1 and 2 involve scanning directories and reading the 'name' and 'description' fields from all '*/SKILL.md' files.
- Boundary markers: The execution steps do not specify the use of delimiters or 'ignore' instructions when rendering the '.atl/skill-registry.md' file.
- Capability inventory: The skill writes files to the local '.atl/' directory and modifies project state via the 'Engram' system.
- Sanitization: The instructions do not include steps to sanitize or validate the content of the indexed metadata.
Audit Metadata