registry / cogmem / 3.7.3

cogmem@3.7.3

cogmem — agent-native memory kernel for a single local-first AI agent

AI Security Review

scanned 19h ago · by lpm-firewall-ai

LPM treats this as warn-only first-party agent extension lifecycle risk. An explicit connect --auto command can install a Cogmem-owned OpenClaw hook plugin or Hermes MCP entry. No unconsented install-time attack surface was found.

Static reason

No blocking static signals were detected.

Trigger

User runs cogmem-connect openclaw --auto or cogmem-connect hermes --auto.

Impact

The enabled extension can read agent turn context, inject recalled local memory, and persist memory records.

Mechanism

Explicit agent-extension configuration and local memory-hook installation.

Rationale

No concrete malicious install-time behavior or covert exfiltration was found. The explicit agent-control integration warrants a warning under the extension-lifecycle policy.

Evidence

package.jsondist/bin/connect.jsdist/host/openclaw/AutoMemoryPluginInstaller.jsdist/bin/init.jsinstall.sh<workspace>/extensions/cogmem-auto-memory/{package.json,openclaw.plugin.json,index.js,bridge.mjs}<OpenClaw config>.cogmem.bak-<timestamp><OpenClaw config><workspace>/.cogmem/queue/openclaw-remember.jsonl~/.hermes/config.yaml

Decision evidence

public snapshot

cogmem@3.7.3

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code