unclawg-internet

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs extracting and embedding access tokens, machine key secrets, and API keys verbatim in outputs (curl headers, env block, browser callback URL, and saved .env files), which requires the LLM to handle secrets in plaintext and poses an exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly instructs launching Task subagents to fetch and research user-provided websites (references/soul-interview.md, Round 2) and to build client overlay query packs that pull public, user-generated content from platforms like Reddit, Twitter/X, Hacker News and LinkedIn (references/artifact-templates.md), which the agent will read and use to generate personas and drive discovery/actions—therefore untrusted third‑party content can influence tool behavior.