Security audit
Security checks for vulnerabilities and agentic risk
This is a straightforward local OCR skill with disclosed first-run language-data downloads and no evidence of hidden data access, persistence, or destructive behavior.
Before installing, understand that OCR runs locally but first use may download Tesseract language data and cache it. For stricter reproducibility, prefer a version that pins tesseract.js and includes a lockfile or validated dependency version.
This markdown file presents all operational instructions, headings, and checklist content in Chinese, which effectively forces a specific language for users reading the skill publication guidance. The policy allows locale constraints only when the skill offers language choice or clearly documents a justified regional limitation, neither of which appears here.
The manifest describes the skill as '100% local, no API key required', which implies operation without remote service interaction. This file explicitly says the Tesseract language data is 'automatically downloaded at runtime', indicating network activity that contradicts the stronger local-only framing in the documentation context.
Lines L73 and L83 present conflicting intent: one claims 'pure local operation', while the first-run note states the user will download Tesseract language data on first use. That is an active contradiction in the documentation about whether network access is involved.
The manifest and README-style description repeatedly describe the skill as '100% local' and 'no API key required', implying no external service interaction. However, the notes explicitly state that the first run downloads language data, which means the skill performs network access at runtime and is not purely local in operation.
The package description states that the skill supports Chinese (simplified/traditional) and English, which creates an explicit language scope in natural-language metadata without clarifying whether users can choose locale behavior or whether other languages are unsupported by design. This is a mild language/locale policy concern because the file presents a fixed language set rather than an opt-in or user-selectable language policy.
Using a caret range for tesseract.js allows installation of different future minor/patch releases, which reduces build reproducibility and can unexpectedly introduce vulnerable or malicious dependency changes. In a local OCR skill, dependency integrity matters because the package processes untrusted files and executes third-party code in the runtime.
"license": "MIT",
"acceptLicenseTerms": true,
"dependencies": {
"tesseract.js": "^7.0.0"
},
"repository": {
"type": "git",
The manifest references tesseract.js without pinning an exact version, while static analysis notes a known advisory affecting some releases. Because the installed version is not fixed, consumers cannot verify whether they are protected, and the skill context increases concern because OCR libraries may parse complex attacker-supplied image data and rely on worker/runtime configuration.
No suspicious patterns detected.