setup

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt tells the agent to read local files (e.g., ~/.claude/CLAUDE.md and vault files) and web content and to pull "example sentences" and other extracted text into generated Context files and outputs, which could cause any secrets present in those sources (API keys, tokens, passwords) to be reproduced verbatim.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill uses the WebFetch tool at runtime to fetch a user-provided website URL (and optional LinkedIn URL), and the fetched page content is directly injected into and used to generate the Context files/prompts, so remote content controls the agent's outputs.