lov-memory-search
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONPERSISTENCE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local knowledge files, which represents an indirect prompt injection surface.\n
- Ingestion points: Workflow steps in SKILL.md specify reading content from user-configured 'memory' and 'distill' directories.\n
- Boundary markers: Absent. The instructions do not mandate the use of delimiters or specific warnings to ignore instructions within the retrieved data.\n
- Capability inventory: SKILL.md authorizes the use of host-provided search, CLI, and browser tools, and scripts/profile_store.py performs file read/write operations.\n
- Sanitization: The skill instructs the agent to use 'safe text matching' for shell characters and to log invalid index lines.\n- [COMMAND_EXECUTION]: The execution boundary in SKILL.md authorizes the agent to use host-provided CLI and browser capabilities, which could be leveraged if the agent follows malicious instructions encountered during file processing.\n- [PERSISTENCE]: The skill implements state persistence for user preferences and records using scripts/profile_store.py to write to a local JSON profile. This script includes a validation check to prevent the storage of sensitive fields like tokens or passwords.
Audit Metadata