repo-to-skill

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

• Third-party content exposure detected (high risk: 0.95). The required workflow clones a user-supplied GitHub/local repository and then reads/copies its README/docs/examples/tests into the skill’s LLM-readable reference files at runtime, so outsider-authored free text from the repo (e.g., README.md, docs/, examples/) can be ingested into the agent context.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly runs git clone on user-specified GitHub repositories at runtime (e.g., git clone https://github.com//.git) and then copies repository docs/code into the skill, so fetched remote content would be injected/used to build prompts and could directly control agent behavior.