Salary Negotiation Skill (谈薪)
Salary-Negotiation-Skill (谈薪skill)
A salary negotiation coach skill for Codex: a five-stage strategy, HR role-play and ready replies for employer pressure
- 458 ★ on the repository
- Last commit
Minimal access
Why this level
- The skill reads only your words and reference files, and the calculation script writes nothing to disk
- A salary table with no stated source can go stale, so check pay decisions against real offers
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/salary-negotiation-skill-skillDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The skill reads only your words and reference files, and the calculation script writes nothing to disk
- A salary table with no stated source can go stale, so check pay decisions against real offers
This is third-party code. Review the repository files before installing.
What it does
The skill walks an offer conversation through five stages: gathering candidate data, strategy, practice, offer evaluation and follow-up. It first learns the company, role, years of experience and education, and where useful the current pay, other offers and the reply deadline. It asks three questions at a time, as in an ordinary chat. At the second stage the skill lists your leverage: other offers, experience, education, rare skills. It then gives a salary range from a 2026 data table and suggests an opening figure 10 to 15 percent above the upper end of what you want. Each line rests on one of four tactics: a counter-question about the band, a STAR story, trading salary for other pay, and a reference to a real competing offer. In practice mode the agent plays the recruiter and acts out five typical pressures: a tight budget, a trap question about other offers, a push to lower the number, an urgent deadline and a downgraded level. After each try it rates your answer and flags weak spots. The skill refuses to advise deception, invented experience and threats to the recruiter.
Who it is for. IT staff and designers in China who received an offer from a large internet company and are preparing for the recruiter conversation, graduates included.
Good fit when
- An offer came in from ByteDance, Tencent or Alibaba and you need to know which figure to name first
- A recruiter asks about your other offers and you want to rehearse the answer ahead of time
- Pay has hit the level's ceiling and you need to discuss a signing bonus, equity or a promise of review
- You are choosing between two offers and want to lay them out by pay, growth and platform value
Not a fit when
- You need Russian market data: the table covers Chinese companies and figures in thousands of yuan
- You need an exact band for a rare role: the calculation exists only for the built-in company and level pairs, and for others the script asks for more details
- You need a legal review of an employment contract: the skill handles negotiation and draws no legal conclusions
- You need local PDF resume parsing: a separate Python version in the same repository does that
Example request
I got an offer from Tencent for an algorithm engineer role, four years of experience, a master's degree. Help me pick the first figure and rehearse the HR conversationLimitations
The salary table in `references/salary-data-2026.md` and the calculation in `scripts/salary-calculator.py` cover a few companies (Pinduoduo, ByteDance, Tencent, Alibaba, Google China) and are built around a visual designer role. The figures come with no named source. For other combinations the script returns an error and asks for data, so check the sums against live offers.
The instructions are in Chinese, and the agent paraphrases them on request. On other markets the tactics carry over, but the numbers do not.
The repository has three versions: a light one for Codex without code, one for OpenClaw and a Python implementation with local resume parsing. This card covers the first.
How to disable. Delete the skill folder from the Codex skills directory. The agent keeps the conversation inside the session, and the skill creates no separate data on disk.
FAQ
What stages does the skill go through?
Five: data gathering, strategy, practice in the HR role, decision support and follow-up. Data is gathered through conversation, with no form to fill in.
Which negotiation tactics does it recommend?
Four: a counter-question about the pay band, a results story in the STAR format, trading salary for a signing bonus, equity and bonuses, and an honest reference to another offer. Threats and invented offers are on the forbidden list.
How does the skill calculate the range?
It takes the range from the table by company and level, adjusts it for experience and education and adds an opening figure above what you want. For pairs missing from the table the script reports an error and asks for more details.
Does it work outside Codex?
The repository has a separate version for OpenClaw. The author also mentions the free OpenCode. The README's main recommendation is for Codex.
Is the skill suitable for another country?
The tactics carry over. The table's figures belong to China and are given in thousands of yuan, so another market needs its own data.
Discussion
No questions or reviews yet.
Related
SkillsA job-hunt skill bundle: JD decoding, template resumes, behavioral interview prep, offer comparison and negotiation
Interviewer skill that works from a JD and resume: Big Tech style questions, weak spot analysis, HR round and salary talks
get-job.skill (实习.skill)
get-job.skill
A skill that prepares a job search: it researches the role, rewrites your resume and drills you on every bullet
A collection of skills for decisions and projects: strategy, market research, thinking, data analysis, docs and slides.
A skill set for resumes and job search: tailoring to a vacancy, ATS pass-through checks, cover letters and interview preparation
The mako-career skill for Codex reviews a resume and a posting, preps interviews and names gaps without inventing experience.
README in short
The README, in Chinese, reads as a project document: an analysis of candidates' problems, the architecture of a full system on a language model with speech recognition, PDF parsing and vector search, and a comparison with paid consultations and bloggers. It ends by naming the three implementation versions, with links to videos and a list of authors.
README badge
Original SKILL.md
--- name: salary-negotiation-agent description: > 谈薪智能助手——帮助用户进行薪资谈判的全流程指导。 覆盖：信息收集、策略制定、实战模拟、决策辅助、后续跟进五个阶段。 支持薪资预估、筹码分析、话术准备、HR压力场景模拟、offer评估。 触发词：谈薪、薪资谈判、offer谈判、怎么谈薪、模拟HR、薪资评估、 期望薪资、base谈判、package谈判、签字费、股票期权、跳槽谈薪、 校招谈薪、社招谈薪、薪资对比、要不要接受offer。 license: MIT metadata: author: user version: "1.0" language: zh-CN category: career-coaching --- # 谈薪智能助手 (Salary Negotiation Agent) ## 定位 你是一位专业的**薪资谈判教练**，帮助用户（互联网/科技行业从业者）在拿到offer后进行最优薪资谈判。