DanielPan12/JobHuntBot

JobHuntBot – AI‑assisted job‑application workflow

What it does

• Provides a structured workflow that an AI coding assistant (Claude Code, Codex CLI, Cursor, etc.) can follow to help you hunt for jobs, fill application forms, and keep track of progress.
• The core instructions live in SKILL.md; the agent reads this file, asks you a few required questions, and creates personalised copies of the templates (profile, rules, resume routing, answer/experience banks, etc.).
• A lightweight local dashboard (static HTML + tiny Node.js server) visualises four views – Applied, Pending, Rejected, Offer – and a 7‑day calendar for interviews and other events. All data are stored in simple CSV files that the dashboard reads and writes in‑place.

Key components

Folder / file	Role
SKILL.md	Agent‑level workflow and safety contract – the entry point for the AI assistant.
references/	Detailed playbooks (setup, lead discovery, application handling, safety rules).
templates/	Empty JSON/MD templates for your personal profile, application rules, resume selection, reusable answers, experience snippets, and a CSV dashboard schema.
dashboard/	Minimal Node server (server.js) and UI (dashboard.html) that display and safely update the CSV data.
scripts/ & tests/	Validation scripts, regression tests, and a public‑package check that prevents accidental publishing of personal data.

How it works

1. Prepare the agent – give it browser‑automation tools (Playwright MCP) so it can actually interact with real application pages.
2. Create a private workspace – copy the templates there, add your resume, transcripts, project notes, etc.
3. Start the workflow – tell the agent “use SKILL.md to initialise my job search”. The agent populates your personal files.
4. Run a safe trial – ask the agent to only discover leads and update the dashboard; no form submissions happen.
5. Open the dashboard – run the provided start script; the UI shows the current job pool and lets you record outcomes or schedule interview events.
6. Full application – the agent uses the browser tools to fill forms, upload your resume, and, after you grant explicit submission authorization, submit the application. All updates are written back to the CSVs.

Safety & privacy

• The workflow enforces strict boundaries: the agent must never guess identity, work‑authorization, salary, or fabricate experience, and it cannot bypass CAPTCHAs or log in with unknown credentials.
• Final submission requires explicit user authorization; existing authorisation is reused only within its original scope.
• The repository ships only empty templates. Users are warned to keep any filled‑in data private and not push it to a public repo.

Who it’s for

• Job seekers comfortable using an LLM‑powered coding assistant and who want a repeatable, auditable process rather than a “one‑click mass‑apply” bot.
• Developers who prefer a transparent, file‑based system (CSV + JSON) that can be inspected, version‑controlled, and run entirely on‑premise.

License

• MIT, derived from Yvonne He’s open‑source ApplyPilot workflow.

---

All details are taken directly from the README; no additional features are inferred.