claude-code

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

• Third-party content exposure detected (high risk: 0.90). The skill's instructions and references explicitly configure and use MCP servers (references/mcp-integration.md) such as Brave Search, Puppeteer, and a GitHub MCP server, and SKILL.md/Slash command examples (e.g., /fix:ci with a GitHub Actions URL and external documentation links like https://context7.com and https://github.com/...) instruct the agent to fetch, read, and act on open/public third‑party content, which could carry untrusted, user‑generated instructions that influence subsequent tool use.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs loading external reference files at runtime and includes a direct fetchable documentation endpoint that can be injected into the model context — e.g. https://context7.com/websites/claude_en_claude-code/llms.txt?tokens=10000 — which meets the criteria of fetching external content used to control prompts.