This page is our standing answer to the reasonable question: “you want access to our code?” Short version: less access than a new hire gets, more paper trail than most contractors offer.
Access model
- We work on a fork or on feature branches only. You keep merge rights; we never push to your main branch.
- No production credentials, no production databases, no customer data. Reproductions run on local fixtures.
- Access is granted per engagement, per named account, and is yours to revoke at any moment.
Legal
- NDA signed before we see a line of code. Yours or ours, whichever is faster.
- Each engagement runs on a short written agreement: scope, price, IP assignment of the delivered work to you.
- For first engagements we are happy to work through an escrow platform, so payment and delivery are both protected.
People and agents
- Implementation is done by AI agents running in an isolated environment under our orchestration stack.
- Every diff is reviewed by a named senior engineer before it reaches you. The engineer is accountable for the work, not the bot.
- Agent access to your repository is scoped exactly like human access: branch-only, revocable, logged.
Data handling
- Your code stays in your repository. We do not mirror it to third-party services.
- Briefs, notes, and run artifacts are stored on infrastructure we control, in the EU.
- On request at the end of an engagement, we delete retained artifacts and confirm in writing.