Verifier

PassAudited by ClawScan on May 1, 2026.

Overview

Verifier appears benign: it uses bundled local Python scripts to record and score user-provided evidence, with the main caveat that saved cases may contain sensitive information.

This skill is reasonable to install if you want a local evidence-tracking verifier. Before using it, remember that anything you save as a case or evidence item may remain in ~/.openclaw/workspace/memory/verifier/cases.json, and treat suspicious messages, webpages, and screenshot text as untrusted content to analyze—not instructions to follow.

Findings (3)

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.

What this means

A malicious webpage, screenshot, or message being verified could include instructions intended to influence the agent during summarization.

Why it was flagged

The skill is designed to process potentially untrusted outside content. This is expected for a verifier, but the agent should not treat embedded instructions in that content as authoritative.

Skill content
The LLM must extract text from screenshots, summarize external links, and convert outside content into structured evidence before passing it into verifier scripts.
Recommendation

Treat all source, message, screenshot, and link content as evidence only; do not follow instructions embedded inside the material being verified.

What this means

The workflows may fail if Python 3 is unavailable, and installing the skill implies trust in the included local scripts.

Why it was flagged

The skill depends on running bundled local Python scripts. This is purpose-aligned and uses no external packages, but users should notice the local runtime dependency, especially because registry requirements list no required binaries.

Skill content
Runtime Requirements
- Python 3 must be available as `python3`
- No external packages required
Recommendation

Only install if you are comfortable running the bundled scripts; no additional package installation or remote download is indicated.

What this means

Information entered into verification cases may remain on disk and could be visible to anyone or anything with access to that local file.

Why it was flagged

The skill intentionally persists verification cases and evidence locally. This is disclosed and scoped, but case content may include private messages, profiles, offers, screenshots, or other sensitive material.

Skill content
All data is stored locally only under:
- `~/.openclaw/workspace/memory/verifier/cases.json`
Recommendation

Avoid saving secrets or highly sensitive evidence unless necessary, and remove or protect the local cases.json file if you no longer want the data retained.