add-golden
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHPROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection (HIGH): The skill is designed to fetch and process external content from URLs, creating a high-risk surface for indirect prompt injection.
- Ingestion points: Phase 1 and 2 in
SKILL.mddescribe fetching and extracting content from user-provided URLs. - Boundary markers: No explicit markers (e.g., XML tags or delimiters) or instructions to ignore embedded commands are used to isolate the fetched content from the agent's logic.
- Capability inventory: The skill uses
TaskCreate,Write, andEdittools. This allows malicious instructions embedded in fetched articles to potentially manipulate the workflow, influence subsequent agents, or corrupt the target dataset. - Sanitization: While
SKILL.mdPhase 7 mentions URL and schema validation, there is no evidence of sanitization to prevent the agents from obeying instructions contained within the document text.
Recommendations
- AI detected serious security threats
Audit Metadata