msw-painter

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The workflow requires capturing the presignedUrl returned by the MCP call and inserting it verbatim into the PUT upload command and the follow-up MCP completion call (fileUrl), which forces the agent/LLM to handle and output a sensitive, signed URL token.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill requires running npm install to fetch the puppeteer package (e.g. https://registry.npmjs.org/puppeteer), which downloads remote Chromium code and at runtime uses puppeteer to load and execute arbitrary injected HTML/JS via page.setContent, so remote-fetched code is both required and executes code in the rendering environment.