yanliudesign/offer-toolkit-skill / job-hunt-skill
job-hunt-skill
Verified批量发现、整理和持续维护求职岗位清单。根据用户的简历、目标方向或种子 JD,搜索 LinkedIn 与公开网页,去重并建立尽量完整的候选池,用证据区分已核验事实和方向性推断,最后生成可搜索、可筛选的单文件 HTML job-hunt-skill 报告。用户说‘帮我批量找岗位’、‘整理职位清单’、‘使用 job-hunt-skill’、‘根据简历搜工作’、‘把这些 LinkedIn 职位做成表格’、‘持续跟踪岗位’时必须使用。只发现和分析岗位;绝不自动申请、处理登录凭据或绕过访问限制。
| File | Kind | Size |
|---|---|---|
| README.md | markdown | 5.7 KB |
| README.zh.md | markdown | 5.5 KB |
| SKILL.md | skill md | 7.3 KB |
| assets/report-spec.md | markdown | 2.6 KB |
| evals/evals.json | config | 1.4 KB |
| examples/cn/job-search-01.png | binary | 403.6 KB |
| examples/cn/job-search-02.png | binary | 600.2 KB |
| examples/cn/job-search-03.png | binary | 568.0 KB |
| examples/cn/job-search-04.png | binary | 442.1 KB |
| examples/cn/job-search-05.png | binary | 372.5 KB |
| examples/en/job-search-01.png | binary | 304.1 KB |
| examples/en/job-search-02.png | binary | 480.2 KB |
| examples/en/job-search-03.png | binary | 500.3 KB |
| examples/en/job-search-04.png | binary | 470.9 KB |
| examples/en/job-search-05.png | binary | 368.4 KB |
| references/evidence-ranking.md | markdown | 2.2 KB |
references/evidence-ranking.md
View on GitHub# Evidence and Ranking Rules
## Evidence hierarchy
Use the strongest available source for each field:
1. Official company job page or visible full job description
2. Public LinkedIn job page
3. Public search-result snippet
4. Title/company inference
Evidence is field-specific. A verified title does not make salary, workplace, posted date, or job status verified.
## Field states
| State | Meaning | Display rule |
|---|---|---|
| `verified` | Directly visible in a public source | Display value |
| `partial` | Visible only in a truncated summary | Display value with limited-evidence note when material |
| `inferred` | Derived from title, company, or adjacent evidence | Label as directional inference |
| `unknown` | No reliable evidence | Leave blank or display `待核验` |
Never convert `unknown` into a plausible value.
## Deep match scoring
Only score roles whose JD requirements are sufficiently visible.
1. Split the JD into Must Have, Nice to Have, Hidden Signals, and Hard Gates.
2. Score each requirement against resume evidence:
- `1.0`: direct evidence with project, scope, result, or concrete detail
- `0.5`: adjacent evidence or unsupported claim
- `0.0`: no evidence
3. Calculate:
```text
Match = 0.60 × MustHave + 0.20 × NiceToHave + 0.20 × HiddenSignalFit
```
Display a range of ±5–8 percentage points, not a decimal point score.
Caps:
- One Must Have scored `0`: maximum 75%
- Two or more Must Haves scored `0`: maximum 55%
- Failed Hard Gate: exclude from main list and summarize the exclusion
- Full JD unavailable: no match percentage; use `未深评`
## Directional ranking
For roles without full JD, rank from visible evidence only:
- Title / level fit: 0–30
- Resume capability overlap: 0–30
- Domain / scope proximity: 0–20
- Freshness: 0–10
- User preference: 0–10
This score is internal. Do not present it as a verified match percentage.
## Reason and gap writing
For deep-reviewed roles:
- `Why it fits` cites two concrete resume facts.
- `Main gap` names the strongest missing or weaker requirement.
For unreviewed roles:
- Label the reason `方向性理由`.
- State which title, domain, or scope signal makes it worth reviewing.
- Label the gap as something to verify, not a known deficiency.
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 |