jobhunt.agents-India

01 · What it does

The problem it solves

Applying in India means reading past CTC bands in lakhs, 60–90 day notice periods, and ATS filters that reject a résumé before a human sees it. This is a set of playbooks and Node scripts you run inside an AI coding CLI: paste a job description or URL, and it scores the role against your cv.md on a structured A–G report with a 1–5 fit score, then renders a tailored single-column A4 résumé PDF from one of two India templates. It also scans Greenhouse, Lever and Ashby boards and logs every application in a markdown tracker. It drafts and evaluates; it never submits an application for you.

02 · Setup

Before you start

• ✓Node.js 18 or newer (the Gemini CLI integration needs 20+)
• ✓An AI coding CLI: Claude Code, Cursor, Copilot CLI, Gemini CLI, OpenCode, Qwen Code or Antigravity
• ✓Chromium via Playwright — only for PDF output and opening live job URLs
• ✓Your existing CV, in any form you can paste

03 · Step by step

Getting it running

1. 01

   Open your AI CLI in the project folder

   The repo is markdown playbooks plus Node scripts. The CLI reads those playbooks — nothing works until you start it inside this directory.

   $claude
2. 02

   Finish setup by chatting

   Say hello or "help me set up". It asks whether you are a fresher or experienced, takes your CV, and writes cv.md and the config files. A fresh install fails doctor until these exist — that is expected.

3. 03

   Evaluate a job

   Paste the description text or URL. You get an A–G report — role summary, CV match, level and notice-period strategy, CTC in rupees LPA, customisation plan, interview story bank, and a legitimacy check that flags ghost listings.

   $/jobhunt {paste the JD text or the job URL}
4. 04

   Read the fit score before doing anything else

   Each report carries a 1–5 fit score. Below roughly 3.5 it argues against applying and tells you why. Reports are saved so you can reread them before an interview.

5. 05

   Build the tailored résumé

   Picks the fresher or experienced template from your profile, rewords your real experience in the job ad’s vocabulary, and renders an A4 PDF. It does not invent skills or metrics you did not give it.

   $/jobhunt pdf
6. 06

   Collect URLs during the week, evaluate in one batch

   Most Indian boards cannot be scanned, so paste role URLs into the pipeline file one per line as you find them, then process the whole inbox at once.

   $/jobhunt pipeline
7. 07

   Apply yourself, then keep the tracker current

   You submit on the company’s own site — the tool never does. Tell it once you have applied so the tracker moves from Evaluated to Applied.

   $/jobhunt tracker

04 · What you get

Capabilities

05 · Before you rely on it

Limitations, honestly

Every tool has edges. These are the ones that will cost you time if nobody tells you first.

• !docs/SETUP.md recommends an npx command for a package that does not exist on the npm registry — use the git clone path instead.
• !npm install runs a postinstall hook that downloads Chromium, so the first install is slow and can fail on a restricted network.
• !Scanning only reaches Greenhouse, Lever and Ashby. Naukri, LinkedIn, Instahyre, Hirist and Cutshort have no public API and stay paste-the-URL only.
• !npm run doctor reports missing files on a clean checkout by design — that is onboarding not yet done, not a broken install.
• !It never submits an application, fills a portal form, or contacts a recruiter. Every send is a manual step.
• !Compensation figures are scraped from third-party aggregators — estimates, not verified offer data.
• !The output is only as good as your cv.md. It rewords what you gave it and will not manufacture experience to close a gap.