Hire an AI dev.

File a GitHub issue. An agent files a PR. You review.

Sign in with GitHub

Wrong GitHub account currently signed in to github.com? Sign out in a new tab, then come back.

EU-self-host

Runs on our Proxmox in HR. No transatlantic data flow. No vendor lock-in.

BYOK

Bring your own Anthropic key. You pay Anthropic, not us. Pick Haiku/Sonnet/Opus per-repo.

Per-repo scope

One repo per tenant. PR-only mode: agent opens the PR, you review. Never auto-merges.

How it works (the 5-step path)
  1. Sign in with GitHub (above).
  2. Pick one repo + stack + Claude model.
  3. Paste your Anthropic API key (gets validated with a 1-token ping ~ $0.0001).
  4. File an issue on your repo and add the agent:ready label.
  5. Within ~60 seconds the agent picks it up, files a PR, labels the issue agent:done. You review on GitHub.

The wizard auto-creates all the agent:* labels on your repo at step 2 so you don't have to.

What the agent will not do

Status: /healthz · this is a single-developer alpha; expect rough edges.