Laand for Claude Code

Review code submissions like you review PRs.

Pull a candidate's micro-project into Claude Code, clone the repo, run their tests and yours, and see exactly where it's strong and where it bends. The way your engineers already work.

Claude Code pulling a candidate's micro-project from Laand, cloning the repo, running the tests and listing weak spots by file

A note from Victor

Small teams don't need another dashboard. You need the hiring to happen where you already work, without giving up the judgement that makes a hire right.

If you're a technical founder, you trust code you've run more than any score. So once you unlock a candidate, Claude Code can clone their micro-project (read-only, one repo, for an hour), run it, and walk you through it like a pull request.

Your agent acts as you, with your role on the team, and sees exactly what you'd see in Laand. Candidates stay anonymous until you unlock them, and anything that spends money stays a click in Laand.

Victor

Set it up

Each person on your team connects their own, in about two minutes. No API keys.

  1. 1
    In your terminal:
    claude mcp add --transport http laand https://laand.me/mcp
  2. 2
    In Claude Code, run /mcp, pick laand and sign in to Laand when asked.

Your first week

What happens after you connect, in the order you'll see it.

  1. 01

    Start a hunt where the work is

    Starting a hunt from the repo the hire will work in means Laand scopes the role from your real code. Ask Claude Code to hire for it, and Laand fills the stack and the hard problems from what it can see.

    Claude Code starting a Talent Hunt in Laand from the repository it's working in
  2. 02

    Clone, run, review

    Ask Claude Code to pull a submitted micro-project. get_submission returns a read-only git clone command for the frozen commit, plus their DECISIONS.md and AI log. Claude Code clones it, runs the tests and lists the weak spots by file.

  3. 03

    How they worked with AI

    Every unlocked submission comes with Laand's score for how they worked with AI: whether they framed the problem well, checked what the AI gave them (every brief may contain a mistake, as real briefs do), and owned the result. Ask Claude Code to compare that with what's actually in the repo.

  4. 04

    Shape the paid work

    Ask your agent to list a role's micro-projects, offer or withdraw each one, and allow or forbid AI on it. Or describe a project you want and Laand drafts it and offers it to talent. The AI rule locks once a candidate accepts, so nobody's rules change mid-project.

    An AI agent listing a hunt's micro-projects, turning AI off on one, and adding the team's own from a sentence
    List, switch off AI, add your own: three requests.
  5. 05

    Unlock, then everything

    Until you unlock a candidate who's sent their work, your agent sees what you see: a made-up name, scores and the opening of Laand's review. Unlock them in Laand (your agent gives you the link and the price) and the full review, their contact details and the work itself arrive in your agent.

How small teams run it

For teams under 20, where the founders still make every hire.

Connect it for every founder who hires

Each person connects their own agent and acts as themselves. Members can read everything and leave their say; owners and admins can also act.

Start every role with one honest paragraph

Tell your agent what's broken and what success looks like in a year. Laand turns it into Target Parameters and asks only for the gaps, and the Deep Analysis is only as good as that paragraph.

Ask "hire or agent?" before you publish

If agents can do most of a milestone today, you may need a different hire, or a smaller one. Better to find out before you spend a stipend.

Let the agent do the reading, keep the deciding

Have it compare submissions, run the code and draft questions. Then meet the person. The tools are built so the decision is always yours.

What it never does

  • Anonymous until you unlock: your agent never sees a name, contact or the work itself before you do.
  • Paying to unlock, inviting to a paid micro-project and marking the hire happen in Laand. Your agent answers with the link.
  • OAuth sign-in, no API keys. Tokens renew hourly; disconnect any agent in Settings › Integrations.
  • Every action from an agent is recorded as yours, "from an AI agent", in Laand and in Slack.

Questions

Does Claude Code see my candidates' names?
Only after you unlock a candidate who has sent their work, exactly as in Laand.
Can it spend money?
No. It can tell you the price and give you the link; a person pays in Laand.
Can it create micro-projects?
Yes. Describe what you want and Laand drafts a fair brief, written or code, and offers it to talent. You can also switch AI on or off for each one.
Is our data used to train AI?
No. Laand's data rules apply to every agent connection. See laand.me/security.
What does it cost?
Nothing to connect. You pay Laand per candidate you unlock, as everywhere else.

Hiring this quarter?

Start a Talent Hunt and meet people who've already done the work.