← Case index
Case file No.038 · npm · stdio
Marrow AI: Control Layer for AI Agents
ai.getmarrow/marrow · v3.9.103
Approval gate and audit trail for AI agent actions: policy check first, proof after. Any MCP client.
Security score
96/100
Strong
Normalized over 100 points of available evidence.
Score breakdown
Known advisories
source: OSV.devNo advisories found in OSV for this package name. (Not proof of safety.)
Metadata change timeline
rug-pull detection- v3.9.102 → v3.9.103Oct 17, 2026
versionchanged version bump"3.9.102" → "3.9.103"
descriptionchanged description text edit"A named yes before your AI agent acts, plus proof the next agent can reuse. Works in any MCP client." → "Approval gate and audit trail for AI agent actions: policy check first, proof after. Any MCP client."
packages.0.versionchanged package version bump"3.9.102" → "3.9.103"
- First snapshot (v3.9.102)Oct 16, 2026
Tool lists: not published by the registry — unknown.
FAQ
auto-generated from the data aboveIs Marrow AI: Control Layer for AI Agents safe?
MCPWatch gives Marrow AI: Control Layer for AI Agents a heuristic security score of 96/100 (strong), based on 100 points of available evidence. This is an automated heuristic, not an audit.
Does Marrow AI: Control Layer for AI Agents have known CVEs or vulnerabilities?
No advisories were found in OSV.dev for the npm package @getmarrow/mcp. That is not proof of safety.
When was Marrow AI: Control Layer for AI Agents last updated?
The latest registry version is 3.9.103, published Oct 17, 2026. The last push to getmarrow/marrow-mcp was on Oct 17, 2026. MCPWatch last checked it on Oct 21, 2026.
What license does Marrow AI: Control Layer for AI Agents use?
GitHub reports the license of getmarrow/marrow-mcp as MIT.