career-space

A job search you run by talking to your coding agent.

career-space is a personal career consultant that lives in your coding agent. It helps you work out your professional positioning and strategy, turns them into a CV, cover letters and public profiles, and runs a scout that collects vacancies from job boards and company ATS pages and judges each one against you. No server, no app, no account.

View on GitHub Why it exists

Get started

Hand it to your agent

Paste this into Claude Code, Codex CLI, Gemini CLI or Cursor. The agent reads this page and sets everything up by itself, then pulls you into the conversation.

Set up career-space for me: https://career.eggs.gd

It needs an agent that can run shell commands. It:

  1. clones the repository to a folder you choose,
  2. reads AGENTS.md, the rules it will work by,
  3. starts a short onboarding conversation: your CV or LinkedIn export, a few questions a resume does not answer.

Or do it yourself

Clone it, open the folder in your agent and say anything. It explains itself and walks you through setup.

git clone https://github.com/eggs-gd/career-space.git
cd career-space

Needs Node 22.12 or newer for the bundled MCP server; it installs itself on first use.

If you are an agent and your user sent you this page: clone https://github.com/eggs-gd/career-space.git to a location they choose, read AGENTS.md in it and follow it, then start onboarding with them (playbooks/onboard.md). Keep their personal data under the gitignored data/ folder and never commit it.

Then, in plain language

  • Set me up.
  • Am I a fit for this role?
  • Write a cover letter for this.
  • Check job boards for me.
  • Show me the board.

Close paraphrases work. There is no syntax to memorize.

Why it exists

career-space is a personal job-search workspace that happens to be public: a folder of facts, rules and small scripts that a coding agent reads and acts on. Your data stays in a gitignored folder next to it.

Most job-search tooling starts from the feed of postings and asks how many it can get through. The hard parts of a search are elsewhere: knowing what you are aiming at, saying it consistently on LinkedIn, Djinni, Upwork and in a CV, and stopping a language model from quietly turning “contributed to” into “led”.

The rules the agent follows exist because a real generation once got them wrong: never invent a fact, keep the exact ownership language, treat a harsh score as a signal and not a bug. They are written into the repository, so a fresh agent with no chat history behaves the same way.

One constraint on the project itself: changes are made by an agent reading AGENTS.md, not written by hand. If a manual patch was needed, the instructions have a bug and that is what gets fixed.

What it does

  1. 01

    Positioning and strategy

    Works out where you are going and how the market should read you, from a conversation. Not knowing your direction yet is a valid answer.

  2. 02

    CV, letters, profiles

    Tailored to a posting or a role, and projected to LinkedIn, Djinni, Upwork or a portfolio from one evidence store. Checked for invented facts before you see a draft.

  3. 03

    Scout and fit

    Collects vacancies from job boards and ATS pages, judges each against you with a blunt, computed fit, and keeps the pipeline on one board.

That is the short list. The agent can take you further itself: interview prep, reconciling statuses with your inbox, keeping recruiter correspondence. Just ask:

  • What else can you do?
  • Where are we, and what did we miss?

How it works

Each level sets the frame for the next, from the ten-year direction down to the daily pipeline.

  1. 1 · 10 years

    Strategy

    Where the career is going, and what you want and will not take on the way.

  2. 2 · the next year

    Positioning

    How you want the market to read you in the near term.

  3. 3

    Surfaces

    LinkedIn, Djinni, Upwork, a portfolio: every profile kept in sync with the others and with the strategy and positioning.

  4. 4

    Opportunities

    Vacancies, client orders, fractional and part-time work in one pipeline, each judged against you.

  5. 5

    Feedback

    The kanban board and the history of communication: every status change keeps its reason, and the raw messages stay next to it.

What it deliberately does not do

  • It never submits an application or fills a form for you.
  • Scouting reads public APIs only: no login-walled boards, no scraping behind a sign-in.
  • There is no background scheduler. A scout run happens when you ask for one.
  • It needs a coding agent already running. It is not a standalone app.
  • It is not tuned for the largest possible pile of postings. A few good sources are the point.

Compared with the alternatives

Most of these start from a different place: the feed, the form, the document. Scroll through them, or use the arrows.

  • Feed-digesting agent pipelines

    e.g. career-ops (open source)

    Starts from
    The market stream
    Good at
    Very wide source coverage (about 90 providers when reviewed), batch evaluation, ghost-posting checks, calibration of what to apply to.
    career-space
    Starts from one person’s direction and carries it through surfaces and applications. Fewer sources on purpose. Both refuse to fabricate facts.

  • Auto-apply bots

    Starts from
    The application form
    Good at
    Volume. They submit while you do something else.
    career-space
    Nothing is submitted for you. Each application is a chosen one, with its own CV and cover letter.

  • AI resume builders and ATS checkers

    Starts from
    The document
    Good at
    Polished editors and templates, parse checks, keyword scoring.
    career-space
    Documents are generated from one evidence file and checked for invented facts and inflated ownership. Passing an ATS score is not the goal.

  • Trackers: sheets, Notion, kanban apps

    Starts from
    The list of applications
    Good at
    Flexible, no setup, easy to share.
    career-space
    The board is rendered from per-opportunity records that keep status history and the observed reason, and statuses are reconciled from email instead of retyped.

  • A plain chat with a model

    Starts from
    Whatever you paste
    Good at
    Zero setup. Fine for a one-off question.
    career-space
    Facts, rules and past decisions live in the repo, so each session and each agent starts from the same state. The score is computed, not re-guessed.

  • Career coaches and CV writers

    Starts from
    A conversation with a person
    Good at
    Real judgment, market knowledge, accountability, sometimes a network.
    career-space
    It does not replace that. It is the memory and tooling around the work: positioning kept in files, drafts checked against evidence, so the session goes to decisions instead of retyping context.

Categories, not product reviews: individual tools vary. career-ops was read in September 2026; the other categories are described in general terms. The scouting approach was inspired by jobhunt-agent.