codebase-vision-report

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted codebase content from local directories or remote Git repositories. Malicious instructions embedded in the analyzed code could influence the agent's final report.
  • Ingestion points: SKILL.md (Step 1 and Step 2) describes reading code from a codebase.
  • Boundary markers: No explicit delimiters or instructions are provided to the agent to distinguish between the codebase content and the analysis prompt.
  • Capability inventory: The skill utilizes npx (Bash), Read, and Write tools, which could be abused if the agent is tricked by the content it reads.
  • Sanitization: There is no evidence of content filtering or sanitization before the codebase is rendered and read.
  • [EXTERNAL_DOWNLOADS]: The skill relies on npx to execute the gitoutput package from the npm registry. This results in the download and execution of external code at runtime without version pinning.
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to run npx gitoutput, which clones repositories and processes local files based on user-supplied arguments.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 10:18 AM
Security Audit — agent-trust-hub — codebase-vision-report