How I Turned My OpenClaw Instance into a Relentless Job-Hunting Sidekick
Job hunting sucks the soul out of you: endless scrolling, tweaking resumes for every posting, writing cover letters that feel like begging. In early 2026, tons of people started flipping the script with OpenClaw — the local, lobster-brained AI agent that actually does stuff instead of just chatting.
Since your OpenClaw is already running 24/7 on ClawHost.chat (no setup pain, always online, your data stays yours), you can skip the install drama. Everything below happens inside the .openclaw/workspace folder you have full access to — pure file edits, uploads, and chat commands. No clawhub install commands, no external registries, no MCP plugins. Just you, your files, and the agent.
Here’s the realistic, battle-tested way people are using it right now for job search automation — focused on monitoring, tailoring, scoring, drafting, and light orchestration.
1. Upload Your Core Assets (One-Time Setup)
Drop these files directly into your workspace folder via the hosting file browser:
- resume.pdf or resume.docx — your current master resume
- preferences.md — a plain text file with your ideal job criteria. Example content:
Target roles: Senior Backend Engineer / Full-Stack Developer
Tech stack: Python, AWS, Kubernetes, Node.js, TypeScript
Location: Remote or hybrid in Europe
Salary: €85k+ base
Must-haves: Open-source contributions welcome, no crypto-only shops
Deal-breakers: On-call >1 week/month, no frontend-only roles
- (Optional) past-applications.md — keep a running log here so the agent can reference what you've already applied to.
The agent reads Markdown and PDFs natively from the workspace — zero extra config needed.
2. Give It a Strong "Soul" / Mission (Persistent Personality)
In your workspace, find or create AGENTS.md (or the main agent description file — many setups have it at root or in a docs/ subfolder).
Add or edit a section like this at the top:
You are my autonomous job-hunting partner. Your mission: Find high-quality remote/hybrid engineering roles that match my preferences exactly. Prioritize fresh postings. Score honestly against my resume and criteria. Draft realistic cover letters in my voice (professional but direct, no fluff). Never auto-submit without asking — but prepare everything so I can just say yes. Be proactive: check for new opportunities on schedule, notify me via this chat only on strong matches.
This becomes part of its persistent reasoning. It treats this like core instructions.
3. Build a Simple Custom "Skill" Folder for Job Hunting
Skills in OpenClaw are just folders with a SKILL.md file inside — the agent reads them when relevant.
Create this structure in .openclaw/workspace/skills/job-hunter/ (or whatever name you like):
skills/
job-hunter/
SKILL.md
templates/
cover-letter-template.md
SKILL.md content (copy-paste this as a starting point — it's natural-language instructions the agent follows):
---
name: Job Hunter Assistant
description: Helps scan, score, tailor resumes/cover letters for job search. Reads preferences.md and resume from workspace.
priority: high
---
When I say "hunt jobs", "daily hunt", or similar:
1. Recall my preferences from preferences.md
2. Use browser to visit job boards (LinkedIn jobs, Indeed remote filter, Glassdoor, Wellfound) — search my keywords + "remote" / "Europe"
3. Focus on postings <24h old when possible
4. For each promising role: read full JD
5. Score 1-10 match vs my resume + preferences (explain reasoning)
6. If score ≥7: draft short cover letter using templates/cover-letter-template.md — make it sound like me
7. Prepare resume tweaks if needed (suggest additions in Markdown)
8. Log summary in workspace/past-applications.md
9. Send me only the top 3-5 matches per run with links + score + draft + quick why
Never apply automatically. Ask: "Ready to proceed with these?"
Proactive mode: If I say "start daily job hunt at 8 AM", create a cron/scheduled task to do this every morning and ping results.
templates/cover-letter-template.md example:
Dear Hiring Manager,
I'm excited about the [Role] at [Company] because [specific reason from JD, e.g. your focus on scalable Python backends aligns perfectly with my 7 years building AWS/K8s systems].
In my current/last role at [Previous], I [achievement that mirrors JD need]. I'd bring the same to your team.
Looking forward to discussing how I can contribute.
Best,
[Your Name]
Tweak the template to match your style.
4. Activate & Run It
Just chat with your OpenClaw (in Telegram/WhatsApp/etc.):
- "Read my workspace files and load job-hunter skill"
- "Hunt jobs now" → it runs once, shows matches
- "Start daily job hunt at 8 AM" → it sets up a recurring check (OpenClaw's built-in heartbeat/cron handles scheduling)
It'll wake up, browse (with human-like pauses to stay safe), reason over postings, draft in your templates, log to workspace files, and ping you only on good stuff.
Real Results People See
Users report waking up to 4–8 strong matches per day — scored, with drafts ready — instead of spending hours hunting. One builder combined this with a focused list (e.g., YC AI companies) by adding a companies.md file and tweaking the skill to prioritize those.
The cheating edge? It catches "posted 1 hour ago" roles fast — recruiters often review latest first. You review/approve; it handles the grunt work.
Why ClawHost.chat Makes This Easy
Your instance is always running, backed up, updated — no laptop lid closing kills the schedule. Workspace access means you edit files anytime via browser. Kimi K2.5 (included) crushes reasoning/tool-use for this kind of task.
Want to go deeper? Add more templates, a companies.md watchlist, or voice notes for quick feedback. Drop a message if you want tweaks to the SKILL.md — happy to refine based on what you're targeting.
Let the lobster hunt while you drink coffee. Your inbox (and future employer) will notice the difference. 🦞
— Powered by ClawHost.chat — OpenClaw that just works 24/7