skills/a-canary/arc-skills/codemap/Gen Agent Trust Hub

codemap

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local system commands git and madge to analyze the project structure and dependencies.
  • [EXTERNAL_DOWNLOADS]: Recommends the installation of the madge package from the npm registry and utilizes npx tsx for script execution.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes arbitrary code and documentation files within a project directory, which could contain malicious instructions if the analyzed repository is untrusted.
  • Ingestion points: The script reads project file names and content from the local file system using fs.readFileSync and git ls-files in scripts/codemap.ts.
  • Boundary markers: The generated Markdown and PlantUML reports do not employ explicit boundary markers or warnings to isolate processed data from agent instructions.
  • Capability inventory: The skill has the ability to write files to the project directory and execute local CLI tools such as git and madge.
  • Sanitization: The script performs basic sanitization of strings used for PlantUML identifiers to prevent syntax errors.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 08:52 PM
Security Audit — agent-trust-hub — codemap