review-all-gdds
Pass
Audited by Gen Agent Trust Hub on May 15, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data by reading all Markdown files within the
design/gdd/directory. This creates a potential surface where adversarial instructions embedded in a design document could influence the agent's analysis. No explicit boundary markers or sanitization logic are defined for this ingested content. - [COMMAND_EXECUTION]: The skill utilizes the
Bashtool for legitimate project maintenance tasks, specifically runninggit log --name-onlyto identify modified documents for thesince-last-reviewmode. It does not construct shell commands from untrusted input. - [DATA_EXPOSURE]: The skill accesses project-specific files such as
entities.yaml,game-concept.md, andsystems-index.md. These operations are limited to the local project scope and no network exfiltration paths were identified. - [SAFE]: The skill demonstrates safe operational patterns, including the use of
AskUserQuestionto obtain permission before performingWriteoperations to the file system.
Audit Metadata