Skip to content

Developer and data

AI agents that use GitLab

UtilitySpaces gives an AI agent scoped access to GitLab through its own encrypted connection, not your password. The agent can call 5 named GitLab tools, and the one that writes waits for a person's approval unless an admin has set the connection to run without asking. Every call is written to a hash-chained audit log with its arguments and its result.

What an agent can do in GitLab

Browse projects, read and file issues, and track merge requests on gitlab.com. An agent calls these by name, and matching is exact.

Every tool a GitLab connection exposes, and whether it writes, which waits for approval by default
ToolWhat it doesWaits for approval by default
List projectsList projects visible to the token.No — read only
Get projectGet a single project by numeric id.No — read only
List issuesList issues in a project.No — read only
Create issueOpen a new issue in a project.Yes — it writes
List merge requestsList merge requests in a project.No — read only

4 read tools, 1 write tool. Generated from the connector catalogue the product runs on, so it cannot list a tool that does not exist.

What teams hand an agent in GitLab

Engineering teams hand an agent the read side of GitLab: projects, issues, and merge requests, listed and summarised in the channel where the conversation is already happening. Create issue is the single write, and it stops for approval unless an admin has let this connection write without asking.

  • Summarise the open merge requests that have waited longest for review in the morning channel digest.
  • Pull the open issues for a project when planning the week, grouped by what they touch.
  • File an issue when a bug is confirmed in the inbox, with the reproduction written up from the thread (after a person approves it)
Before you let it write, or keep out: Have the agent search the existing issues before every Create issue: this connector cannot edit or close issues, so a duplicate that gets through stays open until someone tidies it by hand.

What waits for a person

Reads run on their own. These do not — each one waits in Approvals for an owner or admin unless an admin has set this connection to run writes without asking, and every call toGitLab, approved, rejected or unattended, is logged.

  • Create issue POST

What connecting GitLab takes

GitLab uses an API key you create in GitLab and paste in once. A workspace admin sets the connection up; the fields are:

  • Personal Access Token

Credentials are encrypted and bound to your workspace, never read back, and every outbound call passes one guarded fetch — the connector security posture sets out exactly how, and the security page covers the platform around it.

Where GitLab fits

Give an agent GitLab work by writing a message in the inbox, or put its tools in a scheduled flow. What the run cost is on the cost breakdown.

Other developer and data integrations: GitHub, Sentry. Or see every application agents can connect to.

See it doing your actual work

Open a demo workspace, connect one app, and give an agent a real job. No card, no sales call.

We reply to every message within one business day.

Try the demo