iblai-api-agent-dataset

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

• Third-party content exposure detected (high risk: 0.70). The required workflow includes fetching external public web content (e.g., “Website crawl” / “URL / YouTube / Blackboard” training resources) and/or GitHub branch listings at runtime, which can ingest outsider-authored free text into the agent’s LLM context via the downstream “train”/crawl pipeline.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly fetches and ingests external content at runtime (e.g., listing GitHub branches via https://api.github.com/repos/{owner}/{repo}/branches and adding a YouTube video via https://www.youtube.com/watch?v=dQw4w9WgXcQ), which will be incorporated into the agent's training data and thus directly influence its prompts/responses.