Provider REST API

17 actions

Connect Claude to Buildkite

Read failed builds, job logs and artifacts, and trigger, rebuild, retry or unblock Buildkite builds. Toolspoke puts 17 of its actions behind one MCP endpoint that Claude, Cursor, and Codex all speak.

Connect View pricing

Connection: Provider REST API
Authentication: API access token
Actions exposed: 17
What a call bills: Nothing
Adapter: Maintained by Toolspoke

Connected in three steps

1
Install Buildkite
Open the marketplace in your workspace, add Buildkite to the project your agents work in, and it appears on the gateway immediately.
2
Connect the credential
Authenticate with api 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. Buildkite 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 Buildkite asks for

API 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.

API access tokenRequired: Buildkite → your avatar → Personal Settings → API Access Tokens → New API Access Token. Pick the organization, then tick read_organizations, read_pipelines, read_builds, read_build_logs and read_artifacts. Add write_builds only if this install should trigger, rebuild, retry, cancel or unblock builds.; bkua_…

What Claude can do in Buildkite

17 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: 4Writes
Destructive: 2Destructive

Reads

Fetches data and changes nothing.

list_organizations
List the organizations this token can reach, with the slug every other call takes as org.
list_recent_builds
List the most recent builds across every organization and pipeline this token reaches, newest first. Needs no slug, which makes it the quickest answer to "what failed just now" and the install's connection test - it fails when the token lacks read_builds.
list_pipelines
List the pipelines in one organization, with each pipeline's slug, repository and default branch.
get_pipeline
Retrieve one pipeline, including its steps configuration and branch filters.
list_builds
List builds across every pipeline in one organization, newest first, without their jobs. Filter by state=failed to find what broke, then call get_build on one.
list_pipeline_builds
List builds of one pipeline, newest first, without their jobs.
get_build
Retrieve one build by its number, with every job in it. Each job carries its id, state and exit_status - the id is what get_job_log, tail_job_log, retry_job and unblock_job take.
tail_job_log
Return the last 60 KB of one job's log as plain text. Start here when a job failed - the error is almost always at the end. Use get_job_log when the cause is earlier in the output.
get_job_log
Return one job's whole log as JSON, with the content and its size in bytes. The content is cut at 60,000 characters from the start; use tail_job_log for the end of a long log.
list_build_annotations
List the annotations on one build - the summaries test reporters and pipeline steps write, which often name the failing tests directly.
list_build_artifacts
List the artifacts every job in one build uploaded, with path, size and the job that made each one.

Writes

Creates or updates something on the other side.

create_build
Start a new build of a pipeline for one commit and branch. A write, not destructive - it adds a build and cancel_build stops it - but it runs whatever the pipeline runs, so a pipeline that deploys on this branch will deploy. Use HEAD as the commit for the latest commit on the branch.
rebuild_build
Run a finished build again as a new build with the same commit, branch and environment. Returns the new build.
retry_failed_jobs
Retry every failed job in one build, in place, leaving the jobs that passed alone. Cheaper than rebuild_build when one flaky job broke the build.
retry_job
Retry one failed or timed-out job. Only command jobs that have finished can be retried.

Destructive

Deletes or permanently alters something. Worth granting on purpose.

cancel_build
Cancel a running or scheduled build. Destructive because it cannot be undone - running jobs are stopped part-way, and the build can only be rebuilt from the start, not resumed.
unblock_job
Unblock a block step so the build continues past it. Destructive because block steps are usually a human approval gate - often the one in front of a production deploy - and unblocking cannot be taken back once the next steps start.

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 17 actions above are the whole of it. A call to any other name is refused before it reaches Buildkite 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.
It reaches no further than your credential: Toolspoke holds no access to Buildkite 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 Buildkite: 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 Buildkite. An agent has to ask.
It does not smooth over provider limits: Toolspoke does not retry, queue or back off around Buildkite's own rate limits. A call that Buildkite refuses comes back to the agent as a failed call.

Before you connect it

What can Claude do in Buildkite?

17 named actions: 11 that only read, 4 that write and 2 that delete or permanently alter something. They include list_organizations, list_recent_builds and list_pipelines. 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 Buildkite connector need?

API access token. The connector asks for api access token. 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 Buildkite connector work with Cursor and Codex, or only Claude?

Any client that speaks MCP, and every one of them gets the same 17 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 Buildkite connector not do?

The 17 actions above are the whole of it. A call to any other name is refused before it reaches Buildkite 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. Toolspoke holds no access to Buildkite 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 Buildkite. An agent has to ask. Toolspoke does not retry, queue or back off around Buildkite's own rate limits. A call that Buildkite 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 Buildkite'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 Buildkite?

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 what the call cost in sandbox time and model tokens. 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.

More developer tools connectors

Browse all of them