GitHub

Free and open source · MIT

Run your job search like a launch.

LaunchPad is a job search command center that runs on your own computer. Find roles, find the people who can refer you, ask them, apply, and follow every application to an offer in one place.

Get started Try the live tracker
  • No account
  • No cookies
  • Your data stays on your computer

Scroll to follow one role to an offer ↓

Illustration with sample companies and people. The live tracker is below.

$0
Free, MIT licence
0
Accounts to create
0
Dependencies to install
2
JSON files hold your data
6
Screens in the tracker
11
Job-search skills for your AI agent

The part nobody warns you about

10 roles. 50 conversations. One brain.

Every role wants a referral. Every referral starts with several people. Some accept, fewer reply, and a few refer. Somewhere in there you forget who is waiting on what.

Hand the bookkeeping to an agent

Same 50 conversations. Nothing dropped.

The skills find the people, send what you approved, notice who accepted, draft each ask and follow up once. Every morning Today tells you what needs you.

Run it every day by itself →

Built around three promises.

01

It is yours.

The app is plain Python and HTML with nothing to install. Your data is two JSON files in your own repository, not rows in someone else's database.

02

AI does the legwork.

Finding roles and referrers, drafting each ask, sending what you approved and following up all run as skills your AI agent follows. The tracker still works by hand without it, but the automation is the point.

03

Nothing is sent without your yes.

Connection requests, messages, emails and applications all wait for your approval. Autopilot exists, and it is off until you switch it on.

The real thing, with demo data.

This is the tracker itself, running in your browser with the demo data that ships in the repository. Every name in it is a sample, and nothing you do here leaves this tab.

LaunchPad · live demo
Edits are kept in this tab only. Open full screen

Six screens, one loop

Today, Jobs, People, Companies, Interviews and Insights. Search any of them, or run a command, from one palette.

⌘K

Fit you can read

Every role gets a six-part breakdown: role, skills, level, pay, location and posting.

Referrals as a pipeline

Found, queued, requested, connected, asked. You always know who is waiting on whom.

Insights that stay honest

Funnels for roles and outreach, and a reminder to read the counts before the percentages.

Archive, never delete

Closed and stale items move to an archive folder. Nothing you tracked is thrown away.

On your phone, if you want it

Host your own copy behind a password and it commits each save back to your private repository. The hosting guide walks through it.

Say it. Your agent does the legwork.

Open the folder in an AI coding agent and the skills are already there. Each one reads your own thesis, resume and answer bank.

setup-launchpad · example run
  1. set up LaunchPad
  2. Checks the tracker app that ships in the repository
  3. Asks what roles you want and writes your job thesis
  4. Builds your master resume from the resume you give it
  5. Writes the answer bank that application forms draw on
  6. Offers the daily routine and hosting. Never overwrites what is there.

Running in about a minute.

You need Python 3.9 or later. There are no dependencies and there is no build step.

  1. 1

    Make your own private copy

    Your job search data is committed to your copy, so keep it private.

    Use this template
  2. 2

    Start the tracker

    uv run tracker.py

    No uv? This does the same:

    python3 "LaunchPad App/server.py"

    Then open http://localhost:8787.

  3. 3

    Or look around with demo data first

    uv run Demo/demo_people.py --port 8799
  4. 4

    Connect the agent

    Open the folder in an AI coding agent, such as Claude Code, and say set up LaunchPad. This is what turns the tracker into an automated pipeline.

    Read the setup guide →

Questions people ask.

Is it free?

Yes. LaunchPad is open source under the MIT licence. The skills run in an AI agent that you bring yourself.

Do I need an AI agent to use it?

For the automation, yes. Searching for roles, finding referrers, drafting and sending outreach, filling forms and interview prep are skills an AI agent runs for you. Without one you still have the whole tracker, filled in by hand.

Which AI agents does it work with?

The skills are plain-text playbooks in the repository, so any agent that reads a project folder can follow them. The steps that act for you, such as searching LinkedIn, sending outreach and filling forms, need an agent that can drive a browser and read your email. The guides walk through the setup with Claude Code.

Where does my data live?

In two JSON files, Data/data.json and Data/contacts.json, inside your own copy of the repository. Keep that copy private.

Will it message people or apply for me on its own?

Not unless you say so. Each message and application waits for your approval. Autopilot is opt-in and you choose its rules.

How is this different from a hosted AI job-agent service?

LaunchPad runs on your own computer, costs nothing and is open source, so you can read and change all of it. There is no account, and your applications and contacts never sit on someone else's servers. The trade-off is that you set it up yourself and bring your own AI agent for the automation.

Does it apply to jobs automatically?

It fills the application form from your answer bank and resume, then stops. You read it and press Submit yourself, and the card moves to Applied.

Does it work with LinkedIn?

Yes. One skill searches LinkedIn Jobs in your own logged-in browser, another finds people who could refer you and sends the connection requests you approved, within the daily and weekly limits you set. Referral asks can go by LinkedIn message or by email.

What do the AI skills cost?

LaunchPad itself is free. The skills run in your own AI agent, on whatever plan you already have for it. The tracker by itself needs none.

Does it run on Windows and Linux?

The tracker is plain Python and a web page, so it should run anywhere Python 3.9 does. It is developed and tested on macOS.

How do I get updates?

Keep your private copy pointed at the public repository as an upstream and pull from it. The setup guide shows how.

Can I use it on my phone?

Yes, if you host a copy. The hosting guide puts the tracker online behind a password.

Can it run every day by itself?

Yes. The automation guide sets up a daily routine for the search, the people research and the drafts.

Your next role is a pipeline,
not a pile of tabs.

Use this template Star it on GitHub