search-knowledge
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script memory-manager.py using parameters derived from the active brand configuration to manage memory status and search operations.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data retrieved from external vector databases and knowledge graph servers, which could contain malicious instructions.
- Ingestion points: Data retrieved from connected vector-DB MCP, knowledge-graph servers, and local index files (SKILL.md).
- Boundary markers: None identified to delimit search results from agent instructions.
- Capability inventory: Execution of memory-manager.py via shell; file system reads of brand profiles and SOPs (SKILL.md).
- Sanitization: No specific sanitization or validation of the retrieved content is described.
Audit Metadata