bio-for-cv

Turn a resume, a target JD, and current company context into four recruiter-readable Chinese CV bio lines.

Quick install / 一键安装

Install globally for Codex, Cursor, and Claude Code:

npx skills add StariverKang/bio-for-cv --skill bio-for-cv -g -a codex -a cursor -a claude-code -y

This repository follows the Agent Skills specification and can be installed with the open-source skills CLI.

English

What is bio-for-cv?

bio-for-cv is an Agent Skill for one narrow but high-leverage job-search task: writing the four-line personal summary at the top of a Chinese resume.

It takes three inputs:

the target job description, as text or an accessible link;
the company and role context;
the candidate's current resume or source material.

It then researches the company and role, maps the JD to supported candidate evidence, translates cross-industry experience into the target industry's language, and produces exactly four numbered lines. Each line is 55–120 non-whitespace Unicode characters, with at least 280 characters across the four lines.

Why does a tailored personal summary matter?

The core idea came from an anonymized career-coaching conversation: a resume is rarely consumed like an essay. A recruiter often forms an initial candidate model within seconds, before reading the detailed experience section.

A generic skill list makes the reader do all the interpretation. A strong personal summary does the opposite. It gives the candidate a clear, role-relevant definition first, surfaces the keywords that matter for the target role, and then invites the recruiter to verify those claims in the rest of the resume.

In other words, the summary is not a miniature autobiography. It is an index for the resume:

Who is this candidate in the target market?
What are the two strongest capability clusters?
What methods, tools, metrics, or deliverables make those capabilities credible?
Which parts of the resume should the recruiter inspect next?

The original conversation and candidate materials are not included in this repository. Only the generalized, privacy-safe methodology is published.

What the Skill does

Refreshes company and hiring context from public sources for every run.
Extracts the role's core outcome and highest-priority capability language.
Builds an internal evidence matrix from actions, tools, data, projects, and outputs.
Separates verified facts, capability abstractions, and forward-looking transferable methods.
Preserves the independence of unrelated experiences while extracting shared capabilities.
Adapts vocabulary to the target industry, including GTM, PMO, WBS, Funnel, RAG, CEX/DEX, GMV, Unit Economics, and other role-relevant terms.
Avoids turning availability, year of study, location, or internship duration into generation blockers.
Produces four candidate-centered lines without analysis, citations, headings, or a fifth line.

Four-line funnel

Candidate positioning and relevant background.
The strongest hard-skill workflow and professional deliverables.
A differentiating capability cluster, cross-disciplinary perspective, or evidence-backed working style.
The candidate's highest-value capability intersection with the company and JD, written as existing ability rather than an application explanation.

Usage

After installation, provide your agent with the JD, company/role, and resume, then ask:

Use bio-for-cv to generate the four-line personal summary for this role.
Research the company with current public sources, keep the output in Chinese,
and use only the strongest supported or reasonably transferable capabilities.

The first response is intentionally limited to the four lines. Ask a follow-up if you want the JD keywords, resume evidence, and company sources behind each line.

Installation

Install for one agent:

# Codex
npx skills add StariverKang/bio-for-cv --skill bio-for-cv -g -a codex -y

# Cursor
npx skills add StariverKang/bio-for-cv --skill bio-for-cv -g -a cursor -y

# Claude Code
npx skills add StariverKang/bio-for-cv --skill bio-for-cv -g -a claude-code -y

Manual installation:

git clone --depth 1 https://github.com/StariverKang/bio-for-cv.git
cp -R bio-for-cv/skills/bio-for-cv ~/.agents/skills/bio-for-cv

Agent-specific global locations may differ, so the npx skills add route is recommended.

Privacy and writing boundaries

Resume files, contact details, transcripts, and live company research are never bundled into the Skill.
Public examples are synthetic and anonymized.
Industry terminology may be used to name supported actions, even when the exact term was not present in the source resume.
Adjacent methods may be framed as transferable capability, but specific employment, projects, tools used, metrics, awards, and historical results must have a source.
This is persuasive resume writing, not eligibility screening or background verification.

Development and validation

python3 skills/bio-for-cv/scripts/test_validate_output.py
python3 skills/bio-for-cv/scripts/validate_output.py summary.txt

The validator checks line count, Unicode length, total information density, education repetition, experience narration, employer/JD leakage, punctuation, and semantic misuse of professional terms.

Repository structure

bio-for-cv/
├── README.md
├── LICENSE
└── skills/
    └── bio-for-cv/
        ├── SKILL.md
        ├── references/
        └── scripts/

References

中文

bio-for-cv 是什么？

bio-for-cv 是一个聚焦单一求职任务的 Agent Skill：根据目标 JD、公司信息与候选人履历，生成简历顶部的四条中文个人总结。

它需要三类输入：

JD 正文或可访问链接；
明确的公司与岗位方向；
候选人的当前简历或履历素材。

Skill 会实时研究公司与岗位，把 JD 需求映射到候选人的动作、工具、数据和交付物，将跨行业经历翻译为目标行业熟悉的语言，最后严格输出四条编号文本。每条包含 55–120 个非空白 Unicode 字符，四条合计不少于 280 字符。

为什么适配度高的个人总结很重要？

这套方法的核心观点来自本人在简历辅导经历中沉淀下的方法论与求职潜规则：简历通常不是按文章的方式被逐字阅读。特定行业的 HR 往往会在数秒内先形成一个候选人印象，再决定是否继续阅读详细经历。

普通的技能罗列把解释成本全部留给 HR；高质量个人总结则先替候选人完成定义，用目标岗位熟悉的关键词说明“你是谁、会什么、能完成什么”，再引导 HR 到后文验证这些判断。

因此，个人总结不是缩短版自传，而是你的定位：

候选人在目标行业中是什么类型的人？
最强的两组能力是什么？
哪些方法、工具、指标和交付物能体现专业度？
HR 接下来应该重点查看哪些履历证据？

Skill 会做什么？

每次运行都刷新公司业务、招聘需求与行业背景。
提炼岗位核心目标和 3–5 个高优先级能力词。
建立内部证据矩阵，区分既往事实、能力抽象和前瞻方法。
保持不同履历的独立归因，同时提炼跨经历共性。
根据目标行业选择 GTM、PMO、WBS、Funnel、RAG、CEX/DEX、GMV、Unit Economics 等专业语言。
不把年级、到岗、地点、排班和实习时长变成摘要生成门槛。
严格输出四条候选人中心的个人总结，不附加分析、标题、引用或第五条。

四条固定漏斗

职业定位与相关背景；
最核心硬技能、完整任务链与专业交付；
差异化能力、跨学科视角或有行为证据的工作方式；
候选人与公司及 JD 需求交集最大的能力组合，但仍以候选人已有能力为正文，而不是解释“为什么适合投递”。

使用方式

安装后，把 JD、公司岗位和简历交给 Agent，然后输入：

使用 bio-for-cv 为这个岗位生成四条简历个人总结。
请实时研究公司公开信息，默认输出中文，优先使用最强的直接匹配与可迁移能力。

第一次回复只会包含四条正文。如果需要了解生成依据，再继续询问每条对应的 JD 关键词、履历证据和公司公开来源。

安装

安装到 Codex、Cursor 与 Claude Code：

npx skills add StariverKang/bio-for-cv --skill bio-for-cv -g -a codex -a cursor -a claude-code -y

只安装到单一 Agent：

# Codex
npx skills add StariverKang/bio-for-cv --skill bio-for-cv -g -a codex -y

# Cursor
npx skills add StariverKang/bio-for-cv --skill bio-for-cv -g -a cursor -y

# Claude Code
npx skills add StariverKang/bio-for-cv --skill bio-for-cv -g -a claude-code -y

手动安装：

git clone --depth 1 https://github.com/StariverKang/bio-for-cv.git
cp -R bio-for-cv/skills/bio-for-cv ~/.agents/skills/bio-for-cv

不同 Agent 的全局目录可能不同，推荐优先使用 npx skills add 自动选择正确位置。

隐私与边界

公开示例全部匿名化或虚构化。
可以用行业术语重新命名已有动作，即使原简历没有逐字出现该术语。
可以把相邻方法写成可迁移能力，但具体任职、项目、工具实操、数据、奖项和历史结果必须有来源。
这是候选人利益优先的说服性表达，不是替 HR 进行资格审查或背调。

开发与校验

python3 skills/bio-for-cv/scripts/test_validate_output.py
python3 skills/bio-for-cv/scripts/validate_output.py summary.txt

校验器会检查四行结构、Unicode 字符数、总信息密度、院校重复、经历复述、公司/JD 泄漏、标点以及专业术语的语义误用。

致谢

方法论起点来自一次匿名化求职通话，重点吸收了“HR 数秒阅读”“先定义候选人，再引导验证”“总—分结构”和“不同 JD 使用不同能力语言”等方法论。项目结构与安装方案参考了 Agent Skills 开放规范、OpenAI Plugins、Anthropic Skills 与 skills CLI。

License

MIT

Bio For Cv

At a glance

README