Integrations / GitLab
A coding agent for GitLab merge requests.
Mention @tryreplicas on any GitLab issue or merge request. Replicas starts a coding agent in an isolated cloud VM. From an issue it opens a merge request and links it back. On an MR it pushes the changes you asked for. When a pipeline fails on its MR, it pushes a fix.
Updated
How it works
From GitLab to a reviewed pull request.
- Trigger in GitLab
@tryreplicas on an issue or MR
Replicas picks up the request with its context.
- Isolated cloud VM
Your agent works on your repos
Claude Code, Codex, Cursor or OpenCode runs in its own VM with your environment, and can run tests.
- Result
New MR, or commits pushed to yours
Review the changes with your team before merging.
From GitLab
What the agent can do from GitLab.
| You | Replicas |
|---|---|
| Comment @tryreplicas on an issue | Plans, opens an MR and comments back with a summary |
| Comment @tryreplicas on a merge request | Pushes changes to that MR and replies to you |
| Let a pipeline fail on its MR | Reads the logs, pushes a fix and comments |
| Allowlist bots or users | Responds to them without a mention |
| Open, update, merge or close an MR | Can fire automations for chosen projects |
| Review the MR in Replicas | Approve, merge, rebase or mark it ready |
| Use a GitLab service account | MRs come from a non-human identity |
| Connect your own GitLab account | MRs are attributed to you |
Setup
Connect GitLab in 4 steps.
Full GitLab setup guide- 01
In Replicas, open Organization → Integrations, find GitLab and click Install.
- 02
Authorize Replicas with your GitLab account.
- 03
Add your projects. The connected account needs the Maintainer role so Replicas can add its webhook.
- 04
Mention @tryreplicas on any issue or MR in those projects.
Knowunity uses this
Knowunity keeps repositories in both GitHub and GitLab. One Replicas workspace holds both, so an agent can change the app and backend together and open the matching PR and MR.
Read the Knowunity story- cloud agents launched in July
- 2.8kcloud agents launched in July
- of pull requests produced by cloud agents
- 50%of pull requests produced by cloud agents
- engineering velocity
- 2×engineering velocity
- less time spent fixing bugs
- 2/3less time spent fixing bugs
Other integrations