copy-as-image

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill instructs the agent to paste the full text of a previous assistant response verbatim into a shell here-doc in a single generated Bash command, so any secrets present in that response would be included directly in the LLM's output (exfiltration risk).

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's recommended runtime command runs "npm install" if node_modules is missing, which will fetch and install remote packages (e.g. https://bnpm.byted.org/playwright/-/playwright-1.59.1.tgz and other https://bnpm.byted.org/... resolved URLs from package-lock.json) that are required dependencies and whose code will be executed at runtime (Playwright/other packages), so this is a runtime fetch-and-execute risk.