uinaf-intake

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

• Third-party content exposure detected (high risk: 0.85). The runtime workflow fetches a user-supplied “” (public web content) via bun run intake:find -- "<source-url>", and that outsider-authored text is read/understood to create the summary, so it can flow into the agent’s LLM context for summarization.

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

• Potentially malicious external URL detected (high risk: 0.70). The publish script (scripts/publish-entry.sh) runs "bun install" and then "bun run check:entries" inside the uinaf/intake checkout referenced as https://github.com/uinaf/intake, which will fetch remote packages and execute their code at runtime.