annotate-article-screenshots
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides setup instructions involving a connection to
https://connect.clueso.io/mcp. This is a vendor-owned resource used to enable the skill's primary annotation features. - [INDIRECT_PROMPT_INJECTION]: The skill processes content from user-provided articles, which acts as a potential ingestion point for untrusted instructions.
- Ingestion points: Reads text and images from a user-specified project in SKILL.md.
- Boundary markers: None explicitly specified for the ingestion of user article text.
- Capability inventory: Performs image manipulation (crop, zoom, blur) and article updates through the Clueso MCP toolset.
- Sanitization: The skill includes a mandatory privacy step (Step 4) to detect and blur sensitive data including emails, tokens, and PII found in images.
- [SAFE]: The skill explicitly instructs the agent to identify and obscure sensitive organizational data and credentials, demonstrating a design focused on data protection.
Audit Metadata