source-investigator

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

• Third-party content exposure detected (high risk: 1.00). The skill's workflow (see SKILL.md Step 3 "Clone the exact repository and ref you need" and Step 4 delegating exploration to subagents) requires cloning and reading arbitrary external repositories from public sources, which are untrusted user-generated content that the agent must interpret and that can directly influence its actions and decisions.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs cloning user-specified git repositories at runtime (e.g., https://github.com/owner/repo.git) and using the fetched source in subagents to drive the agent's findings, meaning remote repo content is injected into the agent context and can directly control prompts/outputs.