Provider REST API
Developer toolsConnect Claude to GitLab
Browse projects, triage issues, review merge requests, and check pipelines. Toolspoke puts 14 of its actions behind one MCP endpoint that Claude, Cursor, and Codex all speak.
- Connection
- Provider REST API
- Authentication
- Personal access token
- Actions exposed
- 14
- Cost per call (typical)
- 1 credit
- Adapter
- Maintained by Toolspoke
Connected in three steps
- 1
Install GitLab
Open the marketplace in your workspace, add GitLab to the project your agents work in, and it appears on the gateway immediately.
- 2
Connect the credential
Authenticate with personal access token. Where to get one, and what it has to be able to reach, is the next section.
- 3
Point your agent at the gateway
Give your client one address, https://toolspoke.com/mcp. Claude Code takes it as a command, Claude and Claude Desktop add it as a custom connector, and Cursor, Codex and VS Code each read it from a config file of their own.
.mcp.json{
"mcpServers": {
"toolspoke": {
"type": "http",
"url": "https://toolspoke.com/mcp"
}
}
}One block covers every tool you have installed. GitLab shows up in the client as soon as your policy allows it, and so does everything else you install later.
Where the address goes, per client
- Claude Code
Run it in your project, then /mcp to sign in
claude mcp add --transport http toolspoke https://toolspoke.com/mcp- Claude and Claude Desktop
Settings, then Connectors, then Add custom connector
https://toolspoke.com/mcp- Cursor
~/.cursor/mcp.json, or .cursor/mcp.json for one project
{ "mcpServers": { "toolspoke": { "url": "https://toolspoke.com/mcp" } } }- Codex
~/.codex/config.toml
[mcp_servers.toolspoke] url = "https://toolspoke.com/mcp"- VS Code
.vscode/mcp.json, or the MCP: Add Server command
{ "servers": { "toolspoke": { "type": "http", "url": "https://toolspoke.com/mcp" } } }
What GitLab asks for
Personal access token. You provide it once, when you install the connector. Toolspoke encrypts it at rest and decrypts it only for the length of a single call, and the gateway attaches it to the outbound request itself, so it is never part of the arguments an agent sends.
- Personal access tokenRequired
- GitLab → your avatar → Edit profile → Access tokens. The api scope covers every action here; read_api is enough if you only want the read actions to work. A project or group access token works too and is narrower.
- glpat-…
- GitLab URLOptional
- Leave blank for GitLab.com. For self-managed or GitLab Dedicated, the instance root without /api/v4.
- https://gitlab.example.com
What Claude can do in GitLab
14 actions, each one declared and named by the connector rather than discovered at runtime. A workspace policy grants a person all of them, a hand-picked selection, everything on the read side, everything on the write side, or none.
- Reads
- 11Reads
- Writes
- 3Writes
- Destructive
- 0Destructive
Reads
11Fetches data and changes nothing.
list_projectsFind projects by name, path or description, or list the ones you are a member of. Every other action here needs a project_id, and this is where you get it: use either the numeric id or the full path such as gitlab-org/gitlab, both of which this returns. Set membership true to see only projects you belong to, which is usually what you want on a large instance where an unfiltered search returns thousands of public repositories.
get_projectFetch one project: description, default branch, visibility, clone URLs, web URL, topics and which features are enabled. Call it when you need the default branch name for get_file, or to confirm you have the right repository before writing to it. Takes the numeric id or the URL path from list_projects.
list_issuesList a project's issues with filters for state, labels, milestone, assignee, author and free text. Returns the issue iid - the per-project number shown in the UI as #42 - which every other issue action here takes; the global id in the same response is not interchangeable with it. Use get_issue when you already know the iid and want one issue in full.
get_issueFetch one issue in full: description, state, labels, assignees, milestone, due date, time tracking and web URL. Takes the project-scoped iid from list_issues, not the global id. Use this when you need the actual body; list_issues already carries titles and states.
list_merge_requestsList a project's merge requests, filtered by state, branch, author, reviewer, labels or free text. Returns each one's iid - the per-project number in the UI - plus title, state, merge status and branches. Use get_merge_request for the description, diff statistics and pipeline of a single one. state defaults to opened, which is what a review queue means.
get_merge_requestFetch one merge request in full: description, source and target branches, merge status, approvals state, head SHA, and the pipeline attached to it. Takes the project-scoped iid from list_merge_requests. Read this before commenting with create_merge_request_note, and take diff_refs.head_sha from here if a later action needs it.
list_pipelinesList a project's CI pipelines, newest first, optionally narrowed to one branch, commit SHA or status. Returns each pipeline's id, status, ref, SHA and web URL but nothing about individual jobs - take an id from here and call list_pipeline_jobs to find which job failed. Filter status to failed when you are investigating a broken build.
list_pipeline_jobsList the jobs in one pipeline with each job's name, stage, status, duration, runner and web URL. Call list_pipelines first for the pipeline_id. Set scope to failed to go straight to what broke. This returns job metadata, not job logs - open the job's web_url for the trace.
get_fileRead one file from a repository at a given branch, tag or commit. GitLab returns the content base64-encoded and it is decoded to text for you, so this is only useful for text files - a binary will come back as mojibake. ref defaults to HEAD, which resolves to the project's default branch. Use list_branches if you need a branch name, and get_project for the default branch.
list_branchesList a repository's branches with their latest commit, protection status and whether the token's owner can push to each. Filter with search when a project has hundreds. Use it to confirm a branch name before get_file, or to see what work is in flight alongside list_merge_requests.
get_current_userReturn the account this token belongs to: username, name, numeric id, email and whether it is an administrator. Use it to confirm which GitLab identity an agent is acting as, and to get the numeric user id that create_issue's assignee_ids expects. Takes no arguments and changes nothing, which is why it is the install-time health check.
Writes
3Creates or updates something on the other side.
create_issueOpen a new issue on a project. Only the title is required. Labels are a comma-separated string, and GitLab creates any label that does not already exist, so check list_issues output for the names in use before inventing one. The description is markdown. Returns the created issue including its iid and web URL.
update_issueChange an existing issue: retitle it, rewrite the description, reassign it, adjust labels, or close and reopen it with state_event. Send only what you want changed. Prefer add_labels and remove_labels over labels - labels replaces the whole set, and an empty string strips every label. Closing is reversible from here, which is why this is a write rather than a destructive action.
create_merge_request_notePost a comment on a merge request's main discussion. This is a thread-level note, not a line comment on the diff, and it notifies everyone participating in the review. Get the merge_request_iid from list_merge_requests or get_merge_request first. The body is markdown; GitLab quick actions such as /assign work in it, so keep the text to what you actually mean to say.
What it will not do
Enforced by the gateway rather than left to convention, which is why each of these can be stated flatly.
- It cannot call anything else
- The 14 actions above are the whole of it. A call to any other name is refused before it reaches GitLab rather than forwarded on, and connecting your account does not add to the list: it is fixed by the connector, not discovered at run time.
- Nothing here deletes
- This connector writes to GitLab, but nothing in it deletes or permanently alters anything.
- It reaches no further than your credential
- Toolspoke holds no access to GitLab of its own. Every call carries the credential you stored and nothing besides, so whatever that credential cannot reach, this connector cannot reach either.
- It never hears from GitLab
- Nothing is pushed to it. There is no webhook, no subscription and no polling, so this connector cannot notice by itself that something changed in GitLab. An agent has to ask.
- It does not smooth over provider limits
- Toolspoke does not retry, queue or back off around GitLab's own rate limits. A call that GitLab refuses comes back to the agent as a failed call.
Before you connect it
What can Claude do in GitLab?
14 named actions: 11 that only read and 3 that write. They include list_projects, get_project and list_issues. Nothing outside that list is reachable: the connector declares each operation by name rather than proxying whatever an agent asks for.
What credentials does the GitLab connector need?
Personal access token. The connector asks for personal access token, and optionally gitlab url. Values are encrypted at rest and attached to the outbound request by the gateway, so they are never part of the arguments an agent sends and never reach the audit log.
Does the GitLab connector work with Cursor and Codex, or only Claude?
Any client that speaks MCP, and every one of them gets the same 14 actions. There is a single address, https://toolspoke.com/mcp. Claude Code adds it with claude mcp add --transport http, Claude and Claude Desktop take it as a custom connector in settings, Cursor reads it from .cursor/mcp.json, Codex from ~/.codex/config.toml, and VS Code from .vscode/mcp.json. Each of them signs in to the gateway itself, so there is no key to paste.
What does the GitLab connector not do?
The 14 actions above are the whole of it. A call to any other name is refused before it reaches GitLab rather than forwarded on, and connecting your account does not add to the list: it is fixed by the connector, not discovered at run time. This connector writes to GitLab, but nothing in it deletes or permanently alters anything. Toolspoke holds no access to GitLab of its own. Every call carries the credential you stored and nothing besides, so whatever that credential cannot reach, this connector cannot reach either. Nothing is pushed to it. There is no webhook, no subscription and no polling, so this connector cannot notice by itself that something changed in GitLab. An agent has to ask. Toolspoke does not retry, queue or back off around GitLab's own rate limits. A call that GitLab refuses comes back to the agent as a failed call.
Can I limit which actions an agent can call?
Yes, in two places. The project switches GitLab's actions on and off one at a time, for everyone in the project at once, and the screen groups them by read, write and destructive so turning off everything that deletes is one click. An individual agent key can then be narrowed further, to particular toolkits in a project and to particular actions in a toolkit. Whatever it was granted, a key never reaches a project its owner cannot.
What gets recorded when an agent calls GitLab?
Every attempt, with the agent that made it and the person that agent belongs to, the full request payload, the response payload, the status, the duration, and the credits spent. Values whose key names a secret are masked out before the record is shown to anyone. An operation the connector marks as not retained never has its response body written at all, so the gateway keeps no second copy of what was read.