mem0
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a persistent memory layer that ingests untrusted user input and conversation history, which could contain malicious instructions designed to influence the agent when retrieved.
- Ingestion points: The
scripts/mem0-add.jsscript accepts raw text arguments and JSON-formatted conversation messages via the--messagesflag for memory storage. - Boundary markers: The provided scripts do not include explicit instructions or delimiters to isolate retrieved memories from the primary prompt context.
- Capability inventory: The skill performs file system operations on a local database directory (
~/.mem0/) and makes network requests to OpenAI API endpoints for text embeddings and processing. - Sanitization: No explicit sanitization or validation of the ingested content is visible in the wrapper scripts before the data is passed to the Mem0 library for processing.
- [EXTERNAL_DOWNLOADS]: The skill utilizes the
mem0aipackage and communicates with OpenAI's official API endpoints (text-embedding-3-smallandgpt-4o-mini) for its core functionality.
Audit Metadata