discover-plugins
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use the Bash tool for local directory inspection (
ls plugins/). This is a benign operation used to identify currently installed plugins. - [EXTERNAL_DOWNLOADS]: Recommends installing extensions from the vendor's official marketplace (
ruvnet/ruflo). As these resources originate from the identified skill author, they are considered standard functionality and do not pose an external threat. - [INDIRECT_PROMPT_INJECTION]: The skill features a recommendation engine (
mcp__plugin_ruflo-core_ruflo__guidance_recommend) that ingests user-provided task descriptions. - Ingestion points: User-provided search queries and task descriptions passed to guidance tools.
- Boundary markers: None explicitly defined in the instructions.
- Capability inventory: Local file reading via
Read, directory listing viaBash, and various vendor-specific plugin management tools. - Sanitization: Not applicable as the input is used for semantic matching against a static plugin catalog. The impact is limited to tool discovery recommendations.
Audit Metadata