Back to skill
Skillv1.0.0
ClawScan security
Image OCR Reader · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignMar 3, 2026, 2:23 AM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill's code and instructions are consistent with its stated OCR purpose and request no unexpected credentials or network access.
- Guidance
- This skill appears to do only local OCR with Tesseract and Pillow. Before installing/running: 1) Manually install the tesseract binary and the appropriate language data (chi_sim) as the SKILL.md directs; without the Chinese traineddata, Chinese OCR will fail. 2) Install the Python packages (pytesseract, Pillow) in a controlled environment (virtualenv). 3) Note that --output will write to whatever path you provide; avoid running with paths you don't trust. 4) The code has no networking or secret access, so it won't exfiltrate data, but treat untrusted images cautiously (Pillow vulnerabilities are possible in edge cases). 5) If you plan to use the detailed mode, test it — the code returns confidence values as strings which may cause minor formatting issues when printing; this is a quality issue, not malicious.
Review Dimensions
- Purpose & Capability
- okName/description, SKILL.md, _meta.json, and the included Python code all implement local OCR using Tesseract (pytesseract + Pillow). Required binaries (tesseract) and Python packages match the stated purpose.
- Instruction Scope
- okRuntime instructions and code only read the user-provided image file and optionally write a user-specified output file. There are no references to external endpoints, environment secrets, or unrelated system paths in the SKILL.md or code.
- Install Mechanism
- noteThere is no automatic install spec (instruction-only for system/package installs). The SKILL.md instructs the user to install system tesseract and pip packages manually. This is low-risk but means the user must install external binaries (and Tesseract language data for Chinese) themselves.
- Credentials
- okThe skill requests no environment variables, credentials, or config paths. The needed components (tesseract binary, pytesseract, Pillow) are directly related to OCR functionality.
- Persistence & Privilege
- okThe skill does not request persistent presence (always:false) and does not alter other skills or global agent configuration. It only writes to a user-specified output file if requested.