Coverify
Security checks across static analysis, malware telemetry, and agentic risk
Overview
Coverify appears to be a benign local text-analysis tool, but it runs bundled Python scripts and may save extracted snippets from your inputs in local OpenClaw folders.
This skill is reasonable to install if you want a local heuristic verifier for commitment-token drift. Before installing, be aware that it runs bundled Python scripts, stores model-swap reports locally, and its “proof” or “cryptographic score” language should not be treated as a standalone safety guarantee.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
You are installing and running bundled local scripts from a package whose registry provenance is limited.
Runnable Python code is included even though the registry lacks a source/homepage and install specification. The scripts are visible and purpose-aligned, so this is a provenance/declaration note.
Source: unknown; Homepage: none ... No install spec — this is an instruction-only skill ... Code file presence: 2 code file(s): scripts/commitment_verify.py, scripts/model_swap_test.py
Review the included scripts before relying on the skill, especially if using it in automated governance or research workflows.
Running the model-swap command executes local Python code from the skill package.
The model-swap workflow runs another bundled Python script. It uses an argument list rather than a shell and is central to the stated purpose, so this is expected local execution.
subprocess.run([sys.executable, os.path.join(SCRIPTS_DIR, "commitment_verify.py"), "extract", signal], capture_output=True, text=True)
Run it only in contexts where local Python execution is acceptable, and keep inputs user-directed.
Sensitive input text may leave derived snippets in local result files under ~/.openclaw.
The script persists model-swap reports containing extracted kernels from the provided signal and external kernel input. Those kernels can include commitment-bearing sentences from sensitive text.
RESULTS_DIR = os.path.expanduser("~/.openclaw/governance/model_swap_tests") ... "local_kernel": sorted(local_kernel), "other_kernel": sorted(other_kernel) ... with open(fpath, "w")Avoid using highly sensitive text unless you are comfortable with derived kernels being stored locally, or delete the saved reports after use.
A user might give the tool’s verdict more authority than the implementation warrants.
The wording could encourage users or agents to over-trust the output. The artifacts otherwise describe regex/Jaccard-style commitment comparison, so results should be treated as heuristic evidence rather than absolute proof.
Install it, run it on any signal, get a cryptographic score. No governance harness required.
Use the results as an advisory signal and validate important governance or safety decisions with additional review.