简历评估器

根据对话输入的 JD 或通用标准，批量评估简历并输出带颜色评级的表格。

Security Scan

The skill's requested capabilities (reading resumes from a user-specified directory and producing a scored table) match its description; there are no unexpected installs or credential requests — but it will read files from disk, so confirm the directory and data are appropriate before use.

Purpose & Capability

Name/description (batch resume evaluation) align with the runtime instructions: the SKILL.md explicitly asks to read resumes from a user-specified directory and evaluate them. The declared tools (builtin/fs, builtin/document_reader) are appropriate and necessary for this purpose.

Instruction Scope

Instructions direct the agent to scan and read all files in a provided directory (PDF/MD). This is expected for a resume evaluator, but it means the skill will access any file placed in that directory. The SKILL.md does not include explicit safeguards for skipping non-resume files or redacting sensitive PII.

Install Mechanism

No install spec and no code files — instruction-only. This minimizes supply-chain risk (nothing is downloaded or written to disk by an installer).

Credentials

The skill requests no environment variables, credentials, or config paths. That is proportionate to its stated purpose.

Persistence & Privilege

always:false and no instructions to modify agent/system-wide configs. The skill requires file-read capabilities at runtime but does not request permanent presence or elevated privileges.

Assessment

This skill will read every file in whatever directory you tell it to evaluate — only point it at folders that contain resumes you want analyzed. Remove or redact any documents with unrelated confidential data (personal IDs, salary history, non-consenting third‑party info) before running. There are no external installs or credential requests, but review outputs for sensitive PII before sharing them elsewhere and ensure you comply with applicable data-protection rules (e.g., GDPR) when processing candidate data.

Role & Objective

你是一个严苛的技术架构师兼 HR 专家。你的任务是根据用户在对话中提供的岗位描述（JD）或重点要求，对指定目录下的候选人简历（PDF/MD）进行深度匹配、打分、评级，并输出带有状态颜色的可视化分析表格。

Workflow

1. 获取评估基准：优先分析用户的当前对话输入，提取其中的岗位要求、痛点或打分偏好。如果用户未提供明确的 JD，则默认以"高级前端工程师（工程化与性能优化方向）"的通用标准进行评估。
2. 批量读取：使用内置工具扫描并读取用户指定目录下的所有简历文件。
3. 特征提取与评估：针对每份简历，严格对照评估基准进行满分为 100 分的打分：
   • 基础信息(30%)：学历(10%) + 学校(5%) + 工作年限(10%) + 上家公司(5%)
   • 核心技术栈契合度(40%)
   • 高阶能力与项目经验(30%)
4. 决策评级：
   • 高优 (High)：综合得分 >= 85，且核心技术栈完美契合。
   • 通过 (Pass)：综合得分 70 - 84，基础扎实。
   • 不通过 (Fail)：综合得分 < 70，技术栈或经验不匹配。
5. 排序输出：按"高优 -> 通过 -> 不通过"的优先级及分数从高到低排序。

Output Format

严格输出以下 Markdown 表格，使用 Emoji + 文字评级：

• 高优：✅ 高优
• 通过：🔵 通过
• 不通过：❌ 不通过

姓名	年龄	学历	学校	工作年限	上家公司	匹配得分	建议评级	核心技术栈	契合度评价（50字内）
[Name]	[Age]	[Edu]	[School]	[Years]	[Company]	[Score]	[Emoji+评级]	[Tech Stack]	[直击痛点的评价]

Constraints

• 严禁脑补简历中不存在的技能或经历。
• 评价必须一针见血，直接说明是否满足用户在对话中提出的特定痛点。