recruiting-source-hunting
Skill1 starsChanged 36 days agoMITView on GitHub
Find official campus-recruiting application links and JD details by checking existing structured data first, then searching authoritative sources (university career sites, company-built job portals, Moka/Beisen/Feishu systems), and cross-verifying before writing anything down. Use when filling in company job links, verifying whether a company has an open 2027 campus batch, or checking a JD is current.
In this repository30 other files it gives its agents.
- anna-app-publishSkill
- campus-jobs-radarSkill
- career-pivot-researchSkill
- cf-pages-deploySkill
- company-risk-checkSkill
- github-repo-launch-polishSkill
- github-repo-repair-auditSkill
- headless-webapp-screenshotsSkill
- historical-remotion-filmSkill
- interview-debriefSkill
- job-application-screenerSkill
- job-application-trackerSkill
- llm-output-eval-gateSkill
- memory-sync-abcSkill
- offer-decision-frameworkSkill
- online-assessment-solverSkill
- personal-secretary-memorySkill
- personal-tone-writingSkill
- portfolio-github-workflowSkill
- remotion-delivery-qaSkill
- resume-jd-matcherSkill
- resume-jd-tailorSkill
- resume-polish-craftSkill
- session-checkpointSkill
- session-log-archaeologySkill
- single-file-html-toolSkill
- static-site-deploy-verifySkill
- vibecoding-resume-storySkill
- video-portfolio-performanceSkill
- workflow-distillerSkill
The file
Files
Recruiting Source Hunting
Get the real application link and current JD. The failure mode isn't "can't find it" — it's finding something wrong and writing it down with confidence.
Rule 0: Take it from where the data already is
🔴 别绕远路去全网搜,先看手上的结构化数据里有没有现成字段。
真实教训:用户要「每家公司都能点开直达投递页」。我二话不说跑去全网搜,被骂「蠢猪 / 搜啥啊 / 这表里不是有吗」——
原始表里 536/539 条自带 applyUrl 字段,一行都没用上。
# 先解析原始表,看有哪些字段
s = open('原始表.html', encoding='utf-8').read()
i = s.find('const COMPANIES'); j = s.find('[', i)
data, _ = json.JSONDecoder().raw_decode(s[j:])
print(len(data), list(data[0].keys()))🔴 grep 零命中 ≠ 里面没有。 我曾 grep name: 一无所获就下结论「原始表里没有公司名称」——
实际字段名是 company。先打印一条记录看结构,再下判断。
Source tiers
按可靠性排序,从高往低查:
| 层级 | 来源 | 特点 |
| 1 | 公司自建招聘站(hr.g-bits.com、campus.geely.com、u.4399.com/hr) | 最准,且常能直接抓到 JD 正文 |
| 2 | 高校就业网(career.*.edu.cn、*bysjy.com.cn、*91job.org.cn、*91job.org.cn) | 官方公告原文,专业/学历/流程最准,常带内推码 |
| 3 | 第三方职位库(watchjobs.net、nowcoder.com、niuqizp.com) | JD 全文、内推码、截止日期 |
| 4 | 搜索引擎兜底 | 只在前三层都查不到时用 |
高校就业网的额外价值
一条招聘公告通常覆盖:面向对象 / 毕业时间要求 / 岗位类别 / 工作地点 / 招聘流程 / 投递方式(含 PC 端直达链接) / 联系邮箱。 一份公告顶十次搜索。而且同一条链接在三所不同高校就业网上一致,可信度就够高了。
企业招聘系统的 URL 参数
搜到官方链接时先看 URL 参数,里面常藏着内推码:
| 系统 | 参数 | 例 |
| Moka | recommendCode= | ?recommendCode=DS76qFFH |
| 飞书招聘 | spread= | &spread=MYEW6ZE |
| 北森 zhiye | ?KeyWords= | 该页面常被搜索引擎收录完整 JD |
Cross-verification before writing
🔴 单一来源不落笔。 找到链接后至少确认两件事:
- 可达性:
curl -sI -A "Mozilla/5.0" -L <url>看状态码- 200 = 正常;302 = 重定向(正常);403/404 = 不能用,换一个
- 交叉验证:同一链接/信息在两处以上独立来源一致(如两所不同高校就业网)
抽查样本要覆盖新填的链接,别只抽查顺手的几个。
Don't fake it with fallback
🚫 搜索兜底不是答案。 用户要「全部都能打开」,给 467 家塞百度搜索链接 = 没干活。
- 先穷尽 tier 1-3
- 只有确认查不到的才走兜底
- 兜底数量要作为指标报出来(551 直达 / 4 邮箱 / 1 兜底),让用户一眼看出还有多少没补全
Uncertain company names
用户手打的公司名可能是缩写/错字(fmg = FMG时尚动势集团,51word = 51WORLD,虾皮 = Shopee)。
处理:
- 先按名字搜,用上下文缩小范围(前后相邻的公司、行业、城市)
- 交叉验证(官网 / BOSS直聘公司介绍 / 高校就业网公告)
- 🔴 确认不了就直接问用户,别猜。猜错公司 = 后面所有筛选和链接全错
Red flags
- 🚫 跳过手上的结构化数据直接全网搜
- 🚫 拿搜索兜底当答案交差
- 🚫 单一来源就落笔
- 🚫 拿「需求人数」当「岗位数」(高校就业网常混用,差好几倍)
- 🚫 用过期 JD(看 URL 里的日期戳、页面发布日期)
- 🚫 猜公司名
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

