Skillv1.0.1

ClawScan security

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

Scanner verdict

BenignMar 10, 2026, 12:25 PM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's code, instructions, and requirements are internally consistent with a local resume-parsing and JD-matching tool; it does not request credentials or include network/install mechanisms that contradict its stated purpose.
Guidance: This skill appears to do what it says: local resume text extraction, prompt construction, and structured matching. Before installing/using: 1) Ensure you run the LLM locally or with an on-prem model — the skill builds prompts but relies on whatever model your agent uses; if that model is a remote API, resumes (PII) could be sent externally. 2) Install Tesseract OCR separately and the listed Python packages; run in a controlled environment. 3) Review and test with non-sensitive example resumes to confirm outputs and that the agent does not make network calls. 4) Add a JSON-output validator wrapper in deployment to catch model hallucinations (the scripts instruct the model to 'only return JSON' but that is not enforceable). 5) If handling real candidate data, ensure compliance with privacy rules and consider isolating processing (air-gapped or restricted network) to prevent accidental exfiltration.

Review Dimensions

Purpose & Capability: okName/description (local resume parsing, JD matching) align with included scripts and docs. Scripts implement PDF/DOCX/OCR extraction, build prompts for a local LLM to produce structured JSON, and implement matching rules — all coherent with the stated purpose. No unrelated binaries, env vars, or external services are required in the manifest.
Instruction Scope: noteRuntime instructions tell the agent to extract text from files, build prompts, and pass them to a (local) large model to produce JSON results. The skill does not instruct reading unrelated system files or sending data to external endpoints. NOTE: the claim of 'completely local' depends on the agent's model configuration — if the agent is configured to use a remote API, resume content could be sent externally even though the skill itself does not include network code.
Install Mechanism: noteNo formal install spec (instruction-only), which is low-risk. README/SKILL.md recommend pip installing PyPDF2, python-docx, pillow, pytesseract and installing the Tesseract engine — standard local dependencies. No downloads from arbitrary URLs or packaged installers are present in the manifest.
Credentials: okThe skill declares no required environment variables, credentials, or config paths. All requested actions (file parsing, local OCR, prompt construction) are proportional to the stated purpose.
Persistence & Privilege: okFlags show normal privileges (always:false, model invocation allowed). The skill does not request permanent presence or modify other skills or system-wide settings.