project-audit
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze entire Go repositories, which serves as a large attack surface for indirect prompt injection. Maliciously crafted content within the audited codebase could attempt to influence the audit sub-agents or the final report generation.
- Ingestion points: The skill audits the "complete repository" (whole-project scope) and specifically targets files like
go.mod,Makefile,AGENT.md, and.cursor/rules/*.mdcfor context. - Boundary markers: The skill employs specific architecture profiles (e.g., hexagonal, cqrs) and category briefs to provide structure to the analysis. It also mandates the use of evidence citations (
path:line) to anchor findings to specific code locations. - Capability inventory: The system uses an 'Audit Orchestrator' to launch six read-only reasoning sub-agents and one write-capable 'comment-cleanup' sub-agent using the
composer-2.5-fastmodel to modify files in the working tree. - Sanitization: There is no evidence of explicit sanitization, filtering, or escaping of the project source code before it is passed to the reasoning models.
Audit Metadata