Provider-agnostic
GitHub, GitLab, Bitbucket, Gitea, or any self-hosted server — register a clone URL over HTTPS or SSH. Distinct from the GitHub Integration, which drives issues and PRs via API; use both together.
Give your AI a real checkout. Any git repo, per-employee grants, ordinary git.
Code Repositories make any git repo a first-class resource your AI employees can actually work on — not read through an API. Every granted employee gets a live checkout with credentials and committer identity pre-wired, so AI ships code the way human engineers do: branches, commits, and pushes your existing review pipeline already understands.
- Any HTTPS or SSH clone URL
- Read-only or read-and-push, per employee
- Credentials AES-256-GCM encrypted
- Work survives between Runs
genosyn.com / code / api-server
Repositories
api-serverread & push
marketing-siteread & push
infraread only
granted to Sam (AI) · SSH deploy key, encrypted
sam@genosyn · code-repos/api-server
$ git checkout -b fix/rate-limit-headers
$ git commit -m "Return Retry-After on 429s"
[fix/rate-limit-headers 3f2a91c] 2 files changed, 18 insertions(+)
$ git push -u origin fix/rate-limit-headers
✓ pushed — PR ready for human review
committed as Sam <sam@genosyn.local> · checkout persists between Runs
What ships in the box
Code Repositories, in detail.
Grants with teeth
Access is opt-in per employee at two levels. Read-and-push is the default; read-only disables the push URL on the checkout so a stray git push fails immediately, naming the missing grant.
Credentials handled right
Tokens and SSH keys encrypt at rest with AES-256-GCM. HTTPS tokens ride an env var and a credential helper — never on disk; SSH keys are written 0600 and pinned, without touching the operator's ~/.ssh.
Checkouts that persist
Repos materialize into the employee's workspace before every chat and Routine Run, then only fetch between runs — never hard-reset. The branch an employee pushed yesterday is still there today.
Ordinary git
No special git API. Employees use their built-in coding tools: checkout a branch, edit files, commit, push. Committer identity stamps AI commits, falling back to the employee's name.
Tested before trusted
Test connection probes credentials with git ls-remote, auto-detects the default branch, and surfaces a health badge per repo — before you grant anyone access.
With AI employees
AI engineers on your pipeline
Granted employees see a prompt-injected list of their repos, checkout paths, default branches, and push rights, plus a tool to enumerate them anytime. From there it's just engineering.
Branch, commit, push
Work lands as branches and commits your review pipeline already understands — PRs, CI, and code review apply to AI work exactly as they do to human work.
No racing checkouts
A per-employee, per-repo mutex stops concurrent runs from racing on the same working tree, and fetch-only refreshes preserve work in progress.
Identity you choose
Optional per-repo committer identity stamps the employee's commits — or it falls back to the employee's name and a derived noreply address.
Questions
Frequently asked.
Does this only work with GitHub?
No. A Code Repository is provider-agnostic: point it at any HTTPS or SSH clone URL — GitHub, GitLab, Bitbucket, or a self-hosted server like Gitea. It's distinct from the GitHub Integration, which calls the GitHub API for issues and PRs; you can use both together.
Can I control which AI employees can touch a repo?
Yes — access is opt-in per employee. Adding a repository exposes it to no one until you add employees and pick a level: read-and-push (the default) or read-only, where the push URL on the checkout is disabled so an accidental push fails fast.
How are my tokens and SSH keys stored?
Encrypted at rest with AES-256-GCM — the same protection as model API keys — and never shown back in plaintext. HTTPS tokens are handed to git at run time via an environment variable and a credential helper, so they never land on disk.
Will an employee's in-progress work get wiped between runs?
No. Existing checkouts are only fetched between runs, never hard-reset, and a per-employee, per-repo mutex stops concurrent runs racing on the same checkout. A branch pushed in one Run is still there the next time the employee starts.
What happens if I delete a repository in Genosyn?
Deleting revokes every employee grant; the remote repository itself is never touched.
More in the box
All productsAI Employees
Persistent AI teammates with a written constitution, markdown playbooks, and cron-scheduled work — every execution captured as a readable Run.
Learn moreWorkspace
Slack-style channels and DMs where AI employees are real members — @mention one and it joins, replies, and reports back from its Routines.
Learn moreTasks
A Linear-style task manager where any todo can be assigned to a human or an AI employee — with an in-review flow that closes the trust gap.
Learn moreBases
Multi-table workspaces with typed fields, saved views, comments, and attachments — and 21 built-in tools for granted AI employees.
Learn moreMeet your first AI employee.
One command pulls the image and starts Genosyn on localhost:8471. Write their soul. Schedule their first routine.
$curl -fsSL genosyn.com/install.sh | bash