Interview Evaluation
ReviewAudited by ClawScan on May 14, 2026.
Overview
This instruction-only recruiting skill is mostly coherent, but it asks the agent to use prior conversations/local resume files and persist feedback into skill files, which needs review before using real candidate data.
Before installing, make sure the agent only reads candidate files you explicitly provide or approve, saves reports to a known confidential location, and does not update persistent skill notes without your review.
Findings (3)
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 pull candidate information from previous chats or local resume files into an evaluation workflow without the user clearly selecting the exact source material.
This directs the agent to consult prior conversations and local resume files, which may contain sensitive candidate data, but does not bound the sessions, directories, files, or user approval needed before doing so.
若用户提到"昨天那批名单 / 按名字匹配"，必须先用历史会话和本地简历文件核对正式姓名，再在内部材料里加"姓名匹配说明"表。正式报告中不要出现姓名匹配过程。
Use this only with an explicit folder or named files for each candidate, and require confirmation before accessing prior conversations or local resumes.
Feedback from one evaluation could persist and influence later candidate assessments, potentially carrying forward incorrect, biased, or sensitive information.
The feedback loop instructs persistent updates to the skill’s own knowledge files, which can shape future hiring evaluations and may store sensitive or biased calibration data without review controls.
| 评价偏差 | 在 EVOLUTION.md 记录预测命中率，形成校准数据 | ... | 归因错误 | 补充到 [attribution-errors.md](attribution-errors.md) | ... | 新发现 | 记录到 [EVOLUTION.md](EVOLUTION.md) |
Require explicit approval and review diffs before changing persistent skill files; remove candidate identifiers and keep calibration data separated by workspace or team.
The agent may create a local Markdown file containing candidate evaluation data.
The skill requires creating or delivering a Markdown file. This is purpose-aligned for an interview evaluation deliverable, but it is still a local file-write action.
完成面评后**必须落为 `.md` 文件并交付给用户**（保存到本地或发送到对话），不要只在回复中给出一次性文本。
Choose the destination folder explicitly, avoid overwriting existing files, and treat the generated report as confidential hiring material.