web-atlas-motion
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external design references, curator reviews, and editorial notes retrieved via the web-atlas MCP tool. This introduces a surface where malicious instructions embedded in the design metadata could influence the agent's behavior.
- Ingestion points: Data returned by functions such as
search_curated_referencesandget_page_motionin SKILL.md. - Boundary markers: None specified in the instructions to separate external data from system instructions.
- Capability inventory: The skill allows the agent to propose behaviors and implement working prototypes (code generation).
- Sanitization: No explicit sanitization or filtering of external editorial notes or metadata is mentioned.
- [EXTERNAL_DOWNLOADS]: The skill is configured to connect to an external Model Context Protocol (MCP) server at
https://web-atlas.ayangabryl.workers.dev/mcpto fetch design recordings and interaction data. This endpoint is hosted on infrastructure associated with the skill's author.
Audit Metadata