loop-library

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow can fetch and ingest remote “live catalog” content (catalog.md/catalog.json) from https://signals.forwardfuture.ai/loop-library/ when the user asks for the latest/live catalog, so outsider-authored text in that remote feed could be read and used.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent to read the live catalog at runtime (https://signals.forwardfuture.ai/loop-library/catalog.md and https://signals.forwardfuture.ai/loop-library/catalog.json), and that remote, prompt-shaped content can directly shape the agent's prompts/instructions when the user requests the latest catalog.