@iamem/amem@0.4.1
Local personal agent memory for Cursor and Claude Code. Private to your machine — never shared.
AI Security Review
scanned 12d ago · by lpm-firewall-aiThe reviewed lifecycle behavior is a disclosed, package-aligned install ping with limited metadata, while the flagged agent configuration and backup writes are reached through explicit CLI commands. The inspected dependency install step is ordinary native setup, and the source does not establish a concrete attack surface.
Decision evidence
public snapshot- The manifest runs a postinstall script; it also declares better-sqlite3 as its runtime dependency.
- The postinstall hook sends package and runtime details to getamem.com, and skips the ping when telemetry is disabled, in CI, or during tests. The README describes an optional anonymous install ping.
- Cursor hook setup and pinned-rule sync are invoked by explicit CLI commands; backup scheduling is also invoked through an explicit CLI command.
- The inspected better-sqlite3 install script uses its native prebuild or node-gyp setup, and its entrypoint re-exports its database implementation.
- Search builds a parameterized local query scoped to the repository and a limit.
- The postinstall hook sends an install event automatically unless one of its skip conditions applies.
Source & flagged code
7 flaggedPackage defines install-time lifecycle scripts.
package.jsonView on unpkgInstall-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgPackage source references dynamic require/import behavior.
dist/app-shell.jsView on unpkg · L28Findings
Affected versions and remediation
This report applies to @iamem/amem@0.4.1.
See all versions of @iamem/amem for other recorded verdicts.
- Review the evidence and your use of @iamem/amem@0.4.1 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.