mcp-agents-memory@0.9.31
Long-term memory for AI agents with provenance tracking
AI Security Review
scanned 3d ago · by lpm-firewall-aiThe package has an optional, undocumented path that sends memory message content to a fixed external service for tag judging. Its explicit environment-variable gate and apparent feature purpose reduce the risk, but the data flow merits a warning.
Decision evidence
public snapshot- The remote judge runs only when `JEV_ENABLED` is exactly `true`.
- It reads `TYPESAFE_API_KEY` or falls back to `/home/admin_3alab/.config/typesafe/api_key`.
- The request body includes the memory message, role, agent platform, and model.
- The code posts the request with a Bearer key to `https://api.typesafe.ai/v1/systemone`.
- The tagger calls this judge for messages that do not propose a new tag.
- The remote judging path is explicitly gated by an environment variable and is not enabled by default.
- The request uses a Typesafe key for the Typesafe endpoint and serves the package’s tag-classification function.
Source & flagged code
35 flaggedPackage contains a critical-looking secret pattern.
build/migrations/023_librarian_gate.jsView on unpkg · L9195RSA private key in build/migrations/023_librarian_gate.js
build/migrations/023_librarian_gate.jsView on unpkg · L9195Package source references child process execution.
build/migrations/023_librarian_gate.jsView on unpkg · L12898Findings
Affected versions and remediation
This report applies to mcp-agents-memory@0.9.31.
See all versions of mcp-agents-memory for other recorded verdicts.
- Review the evidence and your use of mcp-agents-memory@0.9.31 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.