repo-system-map

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). At runtime the workflow clones and then reads arbitrary repository-authored text files (README, guidance, manifests, infra/deploy files, and integration tests) and uses that content as LLM evidence to generate the interactive map and its citations.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly clones and uses https://github.com/tamdogood/CodeTerrain at runtime ("Clone the latest default branch of https://github.com/tamdogood/CodeTerrain") and then runs that repo's install/validation/build commands, which fetch and execute remote code required for the publication step.