Skip to content

Repository files navigation

🤖 AI Job Agent

A free, local-first, privacy-conscious autonomous AI job search staff.
Turns AI tools you already have — Hermes, Claude Code, ChatGPT, Cursor, Codex, OpenCode — into your personal job search team.

License: MIT Python 3.10+ Zero External Dependencies Privacy: Local--First


⚡ Above The Fold: Everything You Need To Know

Question Answer
What is this? An autonomous AI job search staff that finds fresh job openings directly from employer career boards, scores them using a transparent rubric, points out real gaps, tailors resumes with zero hallucinations, and delivers a clean morning report every day.
Why should I care? This is NOT a spray-and-pray auto-apply bot. Ten vetted, high-match opportunities with tailored resumes beat 500 spam applications. AI does the tedious research, deduplication, and drafting — you make the final review and submission.
Can I use it? Yes. Built for normal job seekers. No developer experience needed. Works out-of-the-box on Windows, macOS, and Linux.
What do I need? Standard Python 3.10+ (pre-installed or free download) and your existing résumé. No paid API subscriptions required.
What do I do next? Follow the 10-Minute Quickstart below or read START-HERE.md.

🌟 Key Capabilities

  • 🌐 Direct-From-The-Source ATS Discovery: Scans public employer career endpoints directly (Greenhouse, Lever, Ashby, SmartRecruiters, Workable, Recruitee, BambooHR, Personio, Teamtailor) and keeps direct postings as the primary source of truth.
  • 🔭 Optional Broad Tech Discovery: Add --include-freehire to supplement the curated ATS registry with FreeHire's public multi-company tech-job API; direct employer copies win during deduplication.
  • ⚖️ Explainable Rubric Fit Scoring: Evaluates candidate fit across a transparent 6-factor weighted rubric (0–100) with explicit strength and material gap disclosures (no fake LLM match percentages).
  • 🛡️ Strict Anti-Slop / Anti-Fabrication QA: Every claim on a tailored resume or cover letter is independently verified against your candidate ground truth by scripts/claim_check.py. Zero invented metrics, fake skills, or inflated scopes.
  • 🌅 Signature Daily Morning Brief: Produces an executive report summarizing fresh postings, fit scores, rationale, direct employer application links, and tailored files.
  • 🧭 Multi-Track Opportunity Hunting: Supports candidates who legitimately span several lanes (for example AI transformation, product building, growth systems, and fractional advisory) instead of forcing everything into one title.
  • 💼 Consulting / Fractional / Buyer-Signal Research: Normalizes non-job revenue opportunities separately from employment postings, retains source evidence, and recommends the executive/function roles most likely to own the problem.
  • 👤 Evidence-Backed Person Targeting: Researches the actual decision-maker only after an opportunity is qualified; no invented names, emails, or mass outreach.
  • 🗄️ Persistent Application Ledger: Maintains jobs/history.json so you never see or re-process duplicate opportunities across daily runs.
  • 📈 Outcome Feedback Loop: Records applications, replies, interviews, meetings, proposals, offers, hires, wins, and losses so you can see which opportunity lanes and sources actually convert.
  • 🔒 100% Local-First & Privacy Guaranteed: Your resume, contact details, and application materials live strictly in your private local directory and are protected from accidental git commits.

🏗️ Architecture & Dataflow

+-------------------------------------------------------------------------------+
|                             AI JOB AGENT CORE                                 |
+-------------------------------------------------------------------------------+
                                       |
    [1. Candidate Ground Truth]        |        [2. Direct Public ATS Feeds]
    --------------------------         |        ----------------------------
    candidate/MASTER_PROFILE.md        |        Greenhouse  Lever    Ashby
    candidate/SEARCH_PREFERENCES.md    |        Workable    Recruitee  SmartRecruiters
    candidate/VERIFIED_ACHIEVEMENTS.md |        Personio    BambooHR  Teamtailor
    candidate/skills.json              |                    |
                 |                     |                    v
                 |                     |         [3. Ingestion & Normalizer]
                 |                     |         (scripts/ats_engine.py)
                 |                     |                    |
                 |                     |                    v
                 |                     |         [4. Persistent Ledger & Dedup]
                 |                     |         (scripts/jobstore.py)
                 |                     |                    |
                 |                     |                    v
                 |                     |         [5. Live Freshness Verification]
                 |                     |         (scripts/verify_postings.py)
                 |                     |                    |
                 |                     |                    v
                 +---------------------+-------> [6. Hard Exclusion Filter]
                                       |         (scripts/filter_jobs.py)
                                       |                    |
                                       |                    v
                                       |         [7. Rubric Fit Scorer]
                                       |         (scripts/scorer.py)
                                       |                    |
                                       |                    v
                                       |         [8. Select Top 10 Opportunities]
                                       |                    |
                                       +-------> [9. Grounded Résumé Tailoring]
                                       |         (scripts/tailor_engine.py)
                                       |                    |
                                       v                    v
                          [10. Independent Claim QA Check]
                          (scripts/claim_check.py)
                                       |
                                       v
                          [11. Signature Daily Morning Brief]
                          (output/latest_brief.md)

⏱️ 10-Minute Quickstart

Step 1: Clone and Run Setup Wizard

# 1. Clone repository
git clone https://github.com/QaliAI/ai-job-agent.git
cd ai-job-agent

# 2. Run guided setup wizard (creates your local private profile)
python scripts/setup.py

(On macOS/Linux, use python3 scripts/setup.py)

Step 2: Add Your Real Experience

Open candidate/MASTER_PROFILE.md and paste your actual work history, verified skills, and contact details.

Step 3: Run Your Daily Job Search

python scripts/daily_workflow.py

# Optional: broaden tech-role discovery beyond the curated employer registry
python scripts/daily_workflow.py --include-freehire --freehire-days 21 --freehire-country US

In seconds, your morning brief will be generated in output/latest_brief.md and tailored resumes will be saved to output/resumes/!


🌅 Example Morning Brief Report

# 🌅 Daily Job Search Brief — 2026-08-27

**Candidate**: Jordan Taylor (Senior Backend Engineer)
**Target Roles**: Senior Backend Engineer, Distributed Systems Engineer
**Work Mode**: Remote Only | **Target Salary**: $180k – $220k USD

---

## 📊 Executive Summary
* **Total Postings Scanned**: 128
* **Passed Strict Filters**: 42
* **Top Opportunities Identified**: **8**
* **Tailored Resumes Prepared & Verified**: **3**

---

## 🎯 Top Opportunities (Direct Employer Postings)

### 1. [Stripe] — Senior Backend Engineer, Payments Infrastructure
* **Fit Score**: **94/100** | **Confidence**: High | **Recommendation**: **APPLY**
* **Location**: Remote (US) | **Compensation**: $195,000 – $235,000 USD + Equity
* **Source**: Greenhouse Direct · **Status**: Live & Verified
* **Key Strengths**: High-throughput backend architecture, Go/Python, PostgreSQL optimization, Kafka streaming
* **Identified Gaps**: Job mentions Ruby for legacy services; Jordan has Python/Go background (transferable)
* **Tailored Résumé**: `output/resumes/2026-08-27_stripe_senior-backend.md` (Claim QA: 100% Grounded)
* **Direct Application Link**: https://boards.greenhouse.io/stripe/jobs/5928104

---

### 2. [Ramp] — Staff Distributed Systems Engineer
* **Fit Score**: **91/100** | **Confidence**: High | **Recommendation**: **APPLY**
* **Location**: Remote (US) | **Compensation**: $210,000 – $245,000 USD + Equity
* **Source**: Ashby Direct · **Status**: Live & Verified
* **Key Strengths**: Background task execution engine, Redis, PostgreSQL performance tuning
* **Identified Gaps**: None on core requirements
* **Tailored Résumé**: `output/resumes/2026-08-27_ramp_staff-distributed-systems.md` (Claim QA: 100% Grounded)
* **Direct Application Link**: https://jobs.ashbyhq.com/ramp/8831920

🤖 Supported Agent Runtimes

AI Job Agent adheres to the open Agent Skills specification (skills/*/SKILL.md) and standard tool interfaces.

Runtime / Agent Support Level Discovery Method Setup Guide
Standalone CLI Tier 1 (Full) scripts/daily_workflow.py START-HERE.md
Hermes Tier 1 (Full) skills/ + hermes_skills.json runtime/hermes/
Claude Code Tier 1 (Full) .claude-plugin/plugin.json + skills/ runtime/claude-code/
Cursor Tier 1 (Full) .cursor/rules/job-agent.mdc runtime/cursor/
Codex / OpenCode Tier 1 (Full) .codex-plugin/plugin.json runtime/codex/
ChatGPT (Web/App) Tier 2 (Guided) File uploads in Projects / Custom GPT runtime/chatgpt/

⚖️ Feature Comparison

Feature AI Job Agent Mass Auto-Apply Bots Generic Job Boards
Direct Employer Postings ✅ 100% Direct ATS ❌ Scrapes aggregators ❌ Heavy repost duplication
Anti-Hallucination QA ✅ Built-in Claim Check ❌ Unchecked LLM output ❌ N/A
Transparent Rubric Scoring ✅ Explainable 6 factors ❌ Fake "98% match" ❌ Keyword density only
Personal Data Privacy ✅ 100% Local-First ❌ Stored on third-party servers ❌ Sold to recruiters
Free & Open Source ✅ MIT Licensed ❌ Paid subscriptions ❌ Ads & paywalls
Human In The Loop ✅ Human reviews & submits ❌ Spam submissions (banned by ATS) ❌ Manual search fatigue

📅 Daily Automation & Scheduling

  • Windows: Run powershell -ExecutionPolicy Bypass -File .\runtime\scheduling\windows_schedule.ps1 to register an automated 8:00 AM daily background task.
  • macOS: Follow the launchd instructions in docs/MACOS.md.
  • Hermes: Configure native cron scheduling as detailed in runtime/hermes/README.md.

🔒 Privacy Guarantee

Your career history contains sensitive personal data. AI Job Agent is designed from day one to protect your privacy:

  • All files in candidate/, jobs/, and output/ are automatically ignored by .gitignore.
  • No data is ever sent to third-party tracking services or external databases.
  • For more information, read docs/PRIVACY.md.

📚 Documentation & Guides


🤝 Open Source Attribution

AI Job Agent proudly builds upon, adapts, and credits foundational open-source work:

  • JobClaw Skills (jain777/jobclaw-skills, MIT) — ATS query patterns, deterministic hashing, and jobstore design.
  • Job Hunt Skills (Remotivated/job-hunt-skills, MIT) — Evidence-layer verification and anti-fabrication QA concepts.
  • ATS API Reference (ConorsCode/ats-api-reference, MIT) — Public ATS endpoint patterns and production quirk documentation.
  • Resume Tailor (nuin/resume-tailor, MIT) — Grounded master profile formatting and honest framing concepts.
  • RenderCV (sinaatalay/rendercv, MIT) — Resume layout standards and schema compatibility.

See THIRD_PARTY_NOTICES.md for full copyright notices and licenses.


📄 License

Distributed under the MIT License. See LICENSE for details.

About

Free, local-first, privacy-conscious autonomous AI job search staff for Hermes, Claude Code, ChatGPT, Cursor, and Codex.

Topics

Resources

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages