repocache

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

• Third-party content exposure detected (high risk: 1.00). SKILL.md explicitly clones and inspects arbitrary public Git repositories (user-provided GitHub URLs or repos resolved from npm package names) and then reads/searches their files as part of its workflow (see "Part 1: Configuration & Cloning" and "Part 2: Searching"), so the agent consumes untrusted, user-generated third-party content that could include instruction-like text.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill clones user-provided Git repositories at runtime (e.g., https://github.com/org/repo or git@github.com:org/repo.git) and then reads repository files into the agent's search/context to form answers, which means fetched repo content can directly control the agent's prompts/outputs.