prompt-security/clawsec is a security skill suite you add to an AI agent runtime. The project is named ClawSec, is written in JavaScript, and declares an AGPL-3.0 license. The repository was created on 2026-02-05, and its latest release tag is nanoclaw-traffic-guardian-v0.0.2. It ships security skills together with signed advisory information.
The README says the suite targets agent runtimes such as OpenClaw, NanoClaw, Hermes, PicoClaw and their variants, and supports drift detection that catches changes to config files like SOUL.md and to skill artifacts. Drift detection finds the points where the current state differs from the baseline state, while skill integrity verification is the procedure for confirming that installed skills have not been tampered with. Skill artifact verification, configuration change detection, environment audits, an approval procedure for risky installs, and real-time security advisories are bundled together. Signed advisory information is provided as well, so the suite is meant to manage the skill supply chain and agent configuration side by side.
However, the latest release that can be confirmed is only a tag label, with no changelog described. The evaluation scores fill in only momentum 1.6 (on a 0~10 scale), and the novelty, technology and quality scores were not computed. A momentum of 1.6 is a low figure. Stars number 1125, but there are no performance figures, adoption cases, or detection accuracy numbers, and nothing explains how the AGPL-3.0 license terms apply to actual usage. The evidence is thin, so it looks better kept on the watch list than adopted right away.