Skip to content

yanliudesign/offer-toolkit-skill / job-search-skill

linkedin-job-search-skill

Verified

LinkedIn 岗位发现与匹配排序。根据用户给的一份种子 JD 和简历,提取目标岗位画像,自动生成多组 LinkedIn Jobs 搜索,采集、去重并按证据给岗位分层,最终输出可直接投递的 shortlist。用户说‘帮我找工作’、‘在 LinkedIn 搜适合我的岗位’、‘找相似职位’、‘根据简历推荐岗位’、‘job search’、‘find jobs like this’时必须使用。只搜索和推荐,不自动投递、不代替用户登录、不绕过验证码。

FileKindSize
SKILL.mdskill md7.2 KB
evals/evals.jsonconfig1.1 KB
frameworks/ranking-rubric.mdmarkdown2.2 KB

frameworks/ranking-rubric.md

View on GitHub
# Job Ranking Rubric

只对快速筛选前 10 个岗位做深度评分。评分依据是「完整 JD 中的要求」对「用户简历中的证据」,不是 title 或关键词相似度。

## 1 · 拆解要求

从每份 JD 提取:

- Must Have:不满足会显著降低拿面概率的 3-7 条要求
- Nice to Have:加分但通常不构成门槛的要求
- Hidden Signals:ownership、0→1、enterprise、ambiguity、craft、technical depth 等工作方式信号
- Hard Gates:地点、工作授权、证照、语言、on-call、出差等明确门槛

JD 没写的要求不得自行补入。推断出来的信号必须标为「推断」。

## 2 · 单条证据评分

每条要求按简历事实打分:

- `1.0` 直接命中:简历有具体项目/职责,并有规模、结果或可验证细节
- `0.5` 相邻命中:领域相邻、角色深度不足,或只有 claim 没有细节
- `0.0` 未命中:简历没有证据

只看用户给出的简历。不要因为种子 JD 提到某项能力,就假定用户具备它。

## 3 · 总分区间

```text
Base = 0.60 × MustHave + 0.20 × NiceToHave + 0.20 × HiddenSignalFit
```

先按公式算中心值,再给 ±5-8 个百分点的区间。最终显示如 `72-82%`,不要显示 `77.3%`。

上限规则:

- 1 条 Must Have 为 `0`:上限 75%
- 2 条及以上 Must Have 为 `0`:上限 55%
- 任一 Hard Gate 失败:直接 Skip,不进入 Tier A/B
- 完整 JD 不可见:上限 69%,标 `Description unavailable`

## 4 · 新鲜度只作排序,不伪装成匹配

深度匹配分不包含发布时间。两个匹配区间相近的岗位,按以下顺序排序:

1. 发布时间更近
2. Must Have 直接证据更多
3. 风险更少
4. 更符合用户公司/工作方式偏好

这能避免一个刚发布但明显不匹配的岗位仅凭「新」冲到前面。

## 5 · 输出证据

每个入选岗位都输出:

- `Why it fits`:引用 2 条简历事实,不能只复述 JD
- `Main risk`:招聘经理最可能质疑的一点
- `Missing evidence`:简历中看不出的关键要求;没有则写 `None found`
- `Next step`:Apply / Referral first / Deep decode / Portfolio first

不确定的信息写 `Unknown`,不要补全公司规模、薪资、签证政策或岗位状态。
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: 472kUpdated Claude CodeCodexVerifiedInstall
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: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
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 availableStars: 322kUpdated Claude CodeCodexVerifiedInstall