atai-manual-generation-agent

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

• Third-party content exposure detected (medium risk: 0.30). Outsider-authored free text is ingested at runtime via the caller-supplied values.prompt (wired as ${values.prompt} into the bundle connector source.config.default_text) when creating the MGA bundle.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill fetches blueprint documents and performs API calls at runtime against the deployment endpoint https://api.u1.archetypeai.app which resolves blueprints (including connector default_text/prompts and pinned model configs) that directly control the agent's instructions and are relied on for every run, so this runtime external endpoint can change the prompts/instructions the agent executes.