review-shadow-mcp
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements strong safety principles, including 'Secrets never enter chat' and mandatory user confirmation for all write operations. It explicitly delegates sensitive actions like onboarding or authentication to the official platform dashboard (AICP).
- [SAFE]: The workflow uses specific, bounded API calls to the vendor's platform (Speakeasy AI Control Plane). It uses project IDs and slugs provided by the server and includes idempotency keys and version checks to prevent race conditions or unintended mutations.
- [SAFE]: There are no signs of prompt injection, obfuscation, or remote code execution. All tools mentioned (e.g.,
list_projects,decide_shadow_mcp_access,distribute_mcp_to_plugin) are consistent with the skill's stated purpose of managing MCP infrastructure. - [SAFE]: The skill specifically prohibits widening audiences (e.g., to 'Everyone') automatically and requires that the user manually review and confirm every significant state change.
Audit Metadata