mindos-zh
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill recommends installing a CLI tool via npm:
npm install -g @geminilight/mindos. This package is a vendor-owned resource associated with the skill author (GeminiLight). - [COMMAND_EXECUTION]: The skill utilizes the
mindosCLI tool to execute various file and directory management tasks, such as reading, writing, appending, and deleting content within the knowledge base. These operations are intended for structured knowledge management. - [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it is designed to read and process data from external files.
- Ingestion points: Content is ingested into the agent context through file reading and search operations (
mindos file read,mindos search) as described inSKILL.mdandwrite-supplement.md. - Boundary markers: No specific delimiters or safety warnings are required by the instructions to isolate knowledge base content from the agent's primary instructions.
- Capability inventory: The skill has extensive capabilities to modify the workspace, including file creation, modification, and deletion via the
mindosCLI. - Sanitization: The skill does not implement or require sanitization or validation of content retrieved from the knowledge base documents before processing.
Audit Metadata