The Agent Skills Directory

[Indirect Prompt Injection] (MEDIUM): The skill establishes a workflow that ingests untrusted data from external websites via mcp_fetch_fetch and fetch_webpage tools (evidence in SKILL.md). This creates a vector for malicious instructions embedded in external content to influence the agent's behavior. Ingestion points: External documentation and web pages. Boundary markers: Absent; the skill does not define delimiters or ignore-instructions for the fetched content. Capability inventory: The fetched data influences the agent's logic and response content, and the agent has network access. Sanitization: Absent; no validation or filtering is performed on the ingested content.
[Data Exposure & Exfiltration] (LOW): The skill performs network operations to access external, non-whitelisted domains to retrieve documentation.

uncertainty-verification