Serbia Job Search
/install job-search-skill
Job Search — Serbia & Remote
Overview
Three modes:
- Job Digest — Scrape → extract resume → YOU score and output in chat
- Email — Optional. After chat output, send HTML digest via SMTP
- Cover Letter — On demand, YOU write it following strict human-voice rules
File Locations
| File | Purpose |
|---|---|
workspace/resume.pdf |
Resume PDF. User replaces to update CV. |
workspace/roles.txt |
Target job titles, one per line. |
workspace/config.json |
Search, email, and cron settings. |
data/seen_jobs.db |
SQLite — deduplication across runs. Auto-created. |
scripts/scrape.py |
Fetches jobs from LinkedIn, Infostud, HelloWorld.rs. |
scripts/extract_resume.py |
Extracts text from resume.pdf. Prints to stdout. |
scripts/mark_seen.py |
Marks shown jobs as seen. Run after output. |
scripts/send_email.py |
Sends HTML email digest. Only if email.enabled=true. |
Running the Job Digest
Step 1 — Scrape
python scripts/scrape.py
Writes new unseen jobs to data/raw_jobs.json.
Step 2 — Extract resume
python scripts/extract_resume.py
Read the printed resume text carefully — you need it for scoring.
Step 3 — YOU score each job
Read data/raw_jobs.json. For each job score the fit against the resume 1–10.
Scoring guide:
- 9–10: Exceptional. Meets nearly all requirements, strong stack + seniority match.
- 7–8: Strong. Meets most requirements, minor gaps won't block candidacy.
- 5–6: Moderate. Core match but notable gaps or a stretch.
- 3–4: Weak. Some overlap but significant mismatches.
- 1–2: Poor. Little alignment.
Be specific. Reference actual skills/experience from the resume against actual
requirements in the job. Use the tags field where descriptions are absent.
Only include jobs at or above config.json → search.min_score_to_include.
Step 4 — Output in chat
Sort by score descending, grouped by source. Format each job as:
🟢 [9/10] Strong fit
Senior .NET Developer
Luxoft Serbia · Belgrade
Tags: .NET, Azure, SQL
🔗 https://...
↳ [your 2-3 sentence justification]
🟢 = score 8–10 · 🟡 = 6–7 · 🔴 = below 6
Top summary line: "Found X jobs — Y 🟢 strong · Z 🟡 good · W 🔴 moderate"
Step 5 — Mark seen + optional email
Always run after outputting results:
python scripts/mark_seen.py
Then check config for email:
python scripts/send_email.py
If email.enabled is false in config.json, the email script prints "Email disabled"
and exits cleanly — no error, nothing to do.
If email is enabled, it sends the HTML digest automatically.
Setting Up Daily Cron (Automation)
When the user says "set up daily job search" or "schedule this" or "run it every day at [time]":
-
Read
workspace/config.jsonto get the preferred time and timezone. Default is 09:00 Europe/Belgrade unless the user specifies otherwise. -
Update config.json to set
cron.enabled = trueand the user's preferred time. -
Register the cron job with OpenClaw:
openclaw cron add \
--name "Daily Job Search" \
--cron "0 9 * * *" \
--tz "Europe/Belgrade" \
--session isolated \
--message "Run the job search"
Adjust the hour (0 9 = 9am) to match the user's chosen time.
- Confirm to the user: "Daily job search scheduled for 9:00 AM Belgrade time. Your OpenClaw gateway must be running for the schedule to fire."
To cancel: openclaw cron list to find the job ID, then openclaw cron delete \x3Cid>
Enabling Email
When the user says "enable email" or "also send results by email":
-
Ask for: Gmail address, Gmail App Password (16-char, from myaccount.google.com/apppasswords), and recipient email (can be the same address).
-
Update
workspace/config.json:- Set
email.enabledtotrue - Fill in
smtp_user,smtp_pass,recipient
- Set
-
Confirm: "Email enabled. Results will be sent to [address] after each job search."
Note: Gmail requires an App Password (not the regular login password). Generate one at: myaccount.google.com/apppasswords
Cover Letter (On Demand)
When the user asks for a cover letter for a specific job:
- Run
python scripts/extract_resume.pyand read the output. - Use the job title, company, and any description the user provided.
- Write the letter yourself following these rules exactly:
Rules — non-negotiable:
- Do NOT open with "I am writing to apply for..." or any variation
- No AI buzzwords: cut "leverage", "passionate", "dynamic", "synergy", "results-driven", "excited to", "I believe I would be a great fit"
- Reference SPECIFIC things from the resume against SPECIFIC requirements in the job
- Vary sentence rhythm — mix short sentences with longer ones, not uniform blocks
- 3–4 paragraphs, under 350 words total
- End with a direct, confident close — not "I look forward to hearing from you"
- Sign off as the user's name (check resume for name)
- Must sound like a thoughtful person wrote it after reading the JD carefully
Output the letter in chat. The user copies and edits it themselves.
Important Rules
- Never auto-apply to any job under any circumstances.
- Never make external LLM API calls FROM SCRIPTS — scoring and cover letters are handled by YOU directly within this conversation session.
- If
workspace/resume.pdfis missing: stop and tell the user. - If
workspace/roles.txtis empty: stop and tell the user. - Email is always optional — skill works fully without it.
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install job-search-skill - 安装完成后,直接呼叫该 Skill 的名称或使用
/job-search-skill触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
Serbia Job Search 是什么?
Searches LinkedIn, Poslovi Infostud, and HelloWorld.rs for jobs matching your target roles. Scores each listing against your resume PDF (1-10 with justificat... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 32 次。
如何安装 Serbia Job Search?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install job-search-skill」即可一键安装,无需额外配置。
Serbia Job Search 是免费的吗?
是的,Serbia Job Search 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
Serbia Job Search 支持哪些平台?
Serbia Job Search 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 Serbia Job Search?
由 Tanish Sidhu(@tanishsidhu)开发并维护,当前版本 v1.0.0。