yanliudesign/offer-toolkit-skill / offer-compare-skill
offer-compare-skill
VerifiedOffer 对比决策器。用户同时拿到两份(或多份)offer 纠结怎么选时,扮演 Senior Career Decision Advisor:结构化对比 comp / 成长 / AI 敞口 / 公司强度 / 团队风险 / 晋升速度 / 生活方式,识别 front-loaded vs long-term upside、resume value、switch-out 难度、hidden risks,最后给一条**明确、有立场、不中性**的推荐(并附「你是 X 类人选 A,你是 Y 类人选 B」的分叉判定)。三步:贴两份 offer → 补 priorities & 现状 → 自动生成并打开一份 HTML Offer Decision Report。是 offer-toolkit 里 JD Skill / Resume Skill / BQ Skill 拿到结果之后的下一环。关键词:offer compare, 选 offer, 两个 offer, offer decision, TC compare, 该去哪家, which offer, 4-year TC, RSU, sign-on, front-loaded, long-term upside, career decision。
| File | Kind | Size |
|---|---|---|
| LICENSE | other | 1.0 KB |
| README.md | markdown | 9.2 KB |
| README.zh.md | markdown | 9.1 KB |
| SKILL.md | skill md | 10.7 KB |
| examples/offer-compare-template.html | other | 60.8 KB |
| frameworks/dimension-rubric.md | markdown | 5.8 KB |
| frameworks/hidden-signals-dictionary.md | markdown | 6.1 KB |
| frameworks/offer-compare-report.md | markdown | 15.8 KB |
| frameworks/risk-taxonomy.md | markdown | 6.0 KB |
| offer-bank/_index.md | markdown | 3.0 KB |
| offer-bank/_template.md | markdown | 2.3 KB |
| offer-bank/anthropic-vs-linear-staff-pd-202607.md | markdown | 3.9 KB |
| offer-bank/microsoft-stay-vs-linear-staff-pd-202607.md | markdown | 5.6 KB |
| offer-bank/msft-vs-linear-vs-anthropic-pd-202607.md | markdown | 5.7 KB |
| prompts/dimension-compare.md | markdown | 4.8 KB |
| prompts/hidden-signals.md | markdown | 4.1 KB |
| prompts/offer-breakdown.md | markdown | 5.2 KB |
| prompts/recommendation.md | markdown | 5.0 KB |
| prompts/risk-analysis.md | markdown | 3.8 KB |
README.md
View on GitHub<div align="center">
[中文](./README.zh.md) · **English**
# ⚖️ Offer Compare Skill
---
**Turn two (or more) offers into one clear, opinionated decision — in 3 steps.**
[](./LICENSE)
[]()
[](https://github.com/yanliudesign/offer-compare-skill/stargazers)
[](https://claude.ai/code)
[]()
[]()
[]()
[]()
</div>
> 📦 Part of the **[offer-toolkit-skill](https://github.com/yanliudesign/offer-toolkit-skill)** — the full job-hunt bundle (Search · JD · Resume · BQ · Compare · Negotiate). Install the bundle to get all six in one shot.
An agent skill for choosing between multiple offers. Modeled after a Senior Career Decision Advisor who is willing to take a side. Drop in two (or more) offers and it hands back an HTML decision report telling you: the honest 4-year TC range for each (with every assumption listed), a 7-dimension side-by-side scoreboard (Comp · Growth · AI Exposure · Company Strength · Manager/Team Risk · Promo Speed · Lifestyle), 5 hidden signals people usually miss (front-loaded vs long-term upside · resume value · switch-out difficulty · uncertainty), the top 4–6 risks per offer, and — most important — **one clear recommendation** (never a neutral "both look great, up to you") with an "If you are X → A / If you are Y → B" fork so a second archetype is also covered. This is not a side-by-side spreadsheet. It's a call.
---
## How it works — just 3 steps
Invoke the skill with anything like "help me compare these two offers" / "两个 offer 该选哪个" / "which offer should I take" and you go through exactly:
1. **Paste both offers** — Company / Role / Level / Location / TC breakdown (Base + Bonus + RSU with 4-year vest schedule + Sign-on) for each. Missing pieces are batched into one follow-up, not drip-fed one at a time.
2. **Answer priorities & context** (optional but strongly recommended) — pick 2–3 priorities (💰 comp · 📈 growth · 🤖 AI · 🏢 stability · 🌍 lifestyle · 🚀 promo speed · 🎯 resume value · 🔁 future mobility) + tell me about visa, burnout, family, current base, deadline.
3. **HTML decision report auto-opens** — a single-file report is generated to `~/Desktop/Claude skills/offer-compare-<A>-vs-<B>-<YYYYMM>.html` and pops open in your browser. Inside: Verdict, Assumptions block, §1 Offer Breakdown, §2 Dimension Compare (7 axes), §3 Hidden Signals (5), §4 Risk Analysis, §5 Recommendation. Bilingual EN / 中文 toggle + Export PDF / Markdown baked in.
Then loop: change an assumption / add a new priority → the skill regenerates the report. Everything gets cached to `offer-bank/` so "the A vs B thing from last time" is a lookup, not a re-paste.
---
## Decision Report · the 5-section framework
Every wizard run ends with a single-file HTML report at `~/Desktop/Claude skills/offer-compare-<A>-vs-<B>-<YYYYMM>.html`. Fixed 5-section spine, plus a Verdict + Assumptions preamble:
| # | Section | What it answers |
|---|---------|-----------------|
| — | **Verdict** | One-line recommendation (**A** / **B** / **walk-away**) + ⭐ conviction score. Never neutral. |
| — | **Assumptions** | Every assumption listed explicitly (RSU vest schedule · FX rate · bonus target · stock volatility). If any is wrong, re-run. |
| **1** | **Offer Breakdown** | Base / Bonus / RSU / Sign-on / 4-year TC **range** / equity risk / company stability, side-by-side table. |
| **2** | **Dimension Compare** | 7-axis scoreboard: 💰 Comp · 📈 Growth · 🤖 AI Exposure · 🏢 Company Strength · 👤 Manager/Team Risk · 🔁 Promo Speed · 🌍 Lifestyle. |
| **3** | **Hidden Signals** | 5 signals people miss: front-loaded vs long-term upside · uncertainty · resume value · switch-out difficulty. |
| **4** | **Risk Analysis** | 4–6 explicit risks per offer, grouped by equity / role / team / market / life. |
| **5** | **Recommendation** | Clear pick + tradeoff explanation + "If you are X → A / If you are Y → B" fork. |
The report supports **EN / 中文 toggle**, **Export to PDF**, and **Export to Markdown**. Spec: [`frameworks/offer-compare-report.md`](frameworks/offer-compare-report.md) · skeleton: [`examples/offer-compare-template.html`](examples/offer-compare-template.html).
---
## Three iron rules
1. **Never neutral.** One clear pick is mandatory, backed by an "If X → A / If Y → B" fork so a second archetype is also covered. "Both look great, up to you" is banned.
2. **Numbers as ranges + explicit assumptions.** 4-year TC is always `$X ~ $Y` (RSU refresh on/off · stock ±30% · bonus min/max). Every assumption is listed in the Assumptions block at the top. Single-point numbers are this tool's biggest credibility killer.
3. **Never fabricate insider signals.** Company signals (team health / manager style / promo speed) may only come from public data + what the user provided, and must be tagged "inferred from ___". Numbers the user didn't give (base, share price, current comp) are marked `[needs user input]`, never guessed.
---
## File Structure
```
offer-compare-skill/
├── SKILL.md # Entry · routing · 3-step flow · 3 iron rules
├── prompts/ # 5 internal flows
│ ├── offer-breakdown.md # §1
│ ├── dimension-compare.md # §2
│ ├── hidden-signals.md # §3
│ ├── risk-analysis.md # §4
│ └── recommendation.md # §5
├── frameworks/ # Reusable frameworks / rubrics
│ ├── dimension-rubric.md # 7-axis scoring rubric + weights
│ ├── hidden-signals-dictionary.md # Detection rules for the 5 hidden signals
│ ├── risk-taxonomy.md # 5 risk categories × common patterns
│ └── offer-compare-report.md # Final HTML report — spec + mandatory footer
├── examples/ # Reference skeleton (no personal data)
│ └── offer-compare-template.html
└── offer-bank/ # Local cache of past decisions
├── _index.md # Index
└── _template.md # Starter template for new decisions
```
---
## How it thinks
- **Two offers is a decision, not a table.** People who paste "here are two offers" usually already have a gut feel — they want someone honest enough to say it out loud. This skill's job is to be that voice, not to hide behind neutrality.
- **The dimension that decides is usually not comp.** Comp dominates the spreadsheet, but the *decisive* dimension is often AI exposure, resume value, or switch-out difficulty. §2 and §3 exist to surface that.
- **4-year TC without a range is a lie.** RSU refresh assumption alone can swing TC by 30–40%. Ranges force the tradeoff into the open.
- **Every offer has hidden risk. Both.** §4 must be filled for both offers — including the recommended one. Otherwise §5 looks like a sales pitch, not a decision.
---
## Related skills
Pairs with these — the full job-hunt loop:
- [offer-toolkit-skill](https://github.com/yanliudesign/offer-toolkit-skill) — the all-in-one bundle (Search · JD · Resume · BQ · Compare · Negotiate)
- [job-description-skill](https://github.com/yanliudesign/job-description-skill) — Job Description decoder + should-I-apply
- [resume-builder-skill](https://github.com/yanliudesign/resume-builder-skill) — Resume Builder & Beautifier (12 print-ready templates)
- [Behavior-question-skill](https://github.com/yanliudesign/Behavior-question-skill) — Behavioral interview / story bank
- [salary-negotiation](https://github.com/yanliudesign/salary-negotiation) — Negotiate the offer you picked here
```
See a dream job → JD Skill (decode · match · should-I-apply)
↓ decide to apply
Resume Skill (tailor + polish)
↓ get the interview
BQ Skill (mine stories · mock interview)
↓ get the offer(s)
Offer Compare (this skill — pick one, opinionated)
↓ decided
Salary Skill (diagnose · script · counter · sign)
```
---
## License
MIT — fork it, remix it, ship your own version.
Created by [Dreameryanyan](https://www.linkedin.com/in/yanliudesign/) · [LinkedIn](https://www.linkedin.com/in/yanliudesign/) · [X](https://x.com/yanliudreamer) · [Xiaohongshu](https://www.xiaohongshu.com/user/profile/5b2afdf311be104ac3c22931)
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |