Recruiting Resume Screening
ReviewAudited by ClawScan on May 14, 2026.
Overview
The skill is mostly a local resume-screening helper, but it also tells the agent to search prior local files/caches and persistently update its own screening rules, so it needs review before use.
Use this only with resumes and JDs you are authorized to process. Provide exact files or a dedicated folder, do not allow broad automatic searching of Downloads or attachment caches unless you have reviewed the scope, delete extracted text outputs after use, and require explicit approval before the agent modifies any skill standards or catalog files.
Findings (4)
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.
The agent could inspect or mix in unrelated local resumes, JDs, or other private files when trying to reconstruct an old screening batch.
This directs the agent to retrieve prior resumes/JDs from session history, local notes, Downloads, and attachment caches without first requiring a user-approved file list or paths.
用户说“重新评估之前的某批简历”时，**不要**直接沿用旧结论，也不要先让用户重传。... 用会话搜索（或本地笔记）找回... 到本机缓存目录优先查找历史附件：`~/Downloads/` 或对话工具的附件缓存目录
Require the user to approve exact directories or files before searching, show matched filenames before reading contents, and prefer a dedicated per-task folder over broad Downloads/cache searches.
Unreviewed feedback could alter future screening behavior, introduce bias, or poison the skill’s criteria across later tasks.
The skill tells the agent to persistently modify the skill’s evaluation standards and catalog files based on conversational feedback.
用户反馈 → 更新本 Skill 的评估标准。新发现的风险模式 → 补充到 [red-flags-catalog.md]。重大变更 → 写入 [EVOLUTION.md]。
Only update skill files after explicit user approval, present a diff before writing, keep per-user calibration separate from the shared skill, and require versioned review for lasting criteria changes.
Resume contents and extraction metadata may remain on disk after the screening task, which matters because resumes contain personal information.
The helper script writes extracted resume text files and a JSON summary to the chosen output directory.
out_path.write_text(text, encoding=...) ... summary_path.write_text(json.dumps(rows, ensure_ascii=False, indent=2), encoding=...)
Use a dedicated temporary output directory, avoid shared folders, and delete extracted text/summary files when no longer needed.
This is normal for PDF extraction, but package versions and provenance are not pinned, so installs may change over time or be affected by supply-chain issues.
The setup guidance asks the user to install unpinned Python packages from the public package ecosystem.
python3 -m pip install --user pymupdf pdfplumber pdfminer.six
Install from trusted package indexes, pin versions in a requirements file, and review dependencies before use in sensitive recruiting workflows.