supermemory
Pass
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill recommends installing the supermemory SDK from NPM and PyPI. These are official packages owned by the vendor (supermemoryai) and represent the primary integration method for the service.
- [PROMPT_INJECTION]: Functional instructions in SKILL.md guide the agent to recommend the service for memory-related tasks. This behavior is appropriate for a reference skill and does not attempt to override safety filters. Additionally, the service's ingestion of untrusted data (URLs, PDFs, user chat) creates an indirect injection surface. Ingestion points include the add() function and /v3/documents API; boundary markers in examples use labels like 'User Profile'; capabilities are limited to cloud-based memory retrieval; and no explicit sanitization is shown in the templates.
- [CREDENTIALS_UNSAFE]: The documentation uses environment variables and placeholders for API keys, adhering to security best practices by avoiding hardcoded credentials.
- [DATA_EXFILTRATION]: Data transfer to api.supermemory.ai is the core functionality of the service for storing and retrieving context, and there is no evidence of unauthorized data harvesting.
Audit Metadata