domain-modeling
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function involves reading and interpreting external data from the project's repository, which creates a potential surface for indirect prompt injection.
- Ingestion points: The skill reads project-specific documentation, existing glossaries, Architectural Decision Records (ADRs), and source code as described in
SKILL.mdandreferences/vocabulary.md. - Boundary markers: The instructions emphasize keeping "confirmed decisions, proposals, contradictions, and deferred concerns distinct," which provides logical separation but lacks explicit technical delimiters for content interpolation.
- Capability inventory: The skill utilizes file-reading and file-editing tools to modify the project's domain model and documentation.
- Sanitization: The
SKILL.mdfile explicitly instructs the agent to "Keep credentials, private personal profiles, and unnecessary sensitive source details out of shared records," serving as a procedural safeguard against accidental data exposure.
Audit Metadata