Skillv1.0.0

ClawScan security

Scientific Drawing · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

SuspiciousApr 16, 2026, 12:30 PM

Verdict: suspicious
Confidence: medium
Model: gpt-5-mini
Summary: The skill appears to implement a legitimate plotting tool, but there are notable mismatches between what it declares and what it actually requires (Python, matplotlib/Pillow, hardcoded Windows output path and fonts), and the provided file contents are truncated so the full behavior isn't verifiable.
Guidance: This skill appears to be a legitimate Python-based plotting tool, but you should be cautious before installing/using it: - Dependencies: The code requires Python plus matplotlib, Pillow (PIL), and numpy. The manifest does not declare these — verify they are installed in a controlled environment before running. - Execution: The skill is intended to generate and run Python drawing scripts. Running unreviewed code can be risky; inspect the full source (note: the provided file contents in the evaluation were truncated) before executing. - Filesystem: The scripts create and write files to D:\Personal\OpenClaw\figures by default. If you don't want files on that path, change the OUTPUT_DIR in the code to a safe location. Verify the agent will not write to sensitive directories. - Platform assumptions: The code assumes Windows-style paths and certain system fonts (SimHei, Microsoft YaHei). On non-Windows systems it may fail or behave unexpectedly; update font/config or run on a matching environment. - Transparency: Ask the author (or update the skill) to explicitly declare required binaries and Python package dependencies and to remove or parameterize hardcoded paths. If you want to proceed: run the code in an isolated environment (e.g., a disposable VM or container) after reviewing the full, untruncated source and after installing the required Python packages. If you need help reviewing the remaining truncated parts of the files, provide the full file contents and I can re-check for hidden network calls or other unexpected behavior.

Review Dimensions

Purpose & Capability: concernThe skill's name/description (scientific drawing using matplotlib/Pillow) matches the included Python code which draws charts. However the manifest declares no required binaries or dependencies, while the code clearly needs Python plus matplotlib, Pillow (PIL), and numpy. That omission is an incoherence: someone implementing this tool legitimately needs those runtime dependencies.
Instruction Scope: noteSKILL.md instructs the agent to generate and execute Python drawing code and to save output to a local path (D:\Personal\OpenClaw\figures\). The instructions do not attempt to read unrelated system credentials or network endpoints. They do, however, mandate executing code and writing files to a hardcoded local directory and rely on system fonts — both are expected for this purpose but should be declared explicitly.
Install Mechanism: concernThere is no install spec despite the code depending on third-party Python packages (matplotlib, Pillow, numpy). The skill is instruction-only for installation, but failing to declare or provide an install step for required libraries is an operational and transparency gap (not directly malicious, but suspicious).
Credentials: concernThe skill requests no environment variables or credentials, which is fine. However, it hardcodes an absolute Windows output path (D:\Personal\OpenClaw\figures) and creates that directory, and it expects Windows system fonts (SimHei, Microsoft YaHei). Those implicit assumptions about the environment (filesystem location and available fonts) were not declared in requires.config and could cause unexpected writes or failures on other systems.
Persistence & Privilege: okThe skill is not marked always:true and does not request elevated persistent privileges. It writes files into a local folder it creates for outputs, which is normal for a drawing tool.