github-codebase-search

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly performs semantic searches of public GitHub repositories and returns file contents (see SKILL.md "Semantic search for public GitHub repositories" and scripts/github-codebase-search.py which calls mcporter/morphmcp.github_codebase_search with owner_repo or github_url), meaning it ingests untrusted, user-generated code from the open web that can directly influence agent decisions.

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

• Potentially malicious external URL detected (high risk: 1.00). The script can fetch and execute remote code at runtime via the bunx invocation "bunx @morphllm/morphmcp@latest" (npm package @morphllm/morphmcp), which would run externally-provided code the skill relies on.