Which is best for fixing a broken build?

CircleCI's server is built for that — its headline use is pulling detailed build-failure logs so the agent can propose a fix in the same conversation, plus finding flaky tests and validating config.

Can Buildkite's server create pipelines?

Yes. Beyond listing and inspecting, it can create and update pipelines, read builds, enumerate clusters and queues, pull job logs, and trigger builds.

Buildkite vs CircleCI

Both of these official MCP servers connect an agent to a CI/CD platform so it can help with pipelines and failures, but they emphasize different parts of the loop. Buildkite's server exposes Buildkite Pipelines and Test Engine data: an agent can enumerate the organization, clusters, and queues, list and inspect pipelines, create and update pipelines, read builds with status and commit info, pull job logs (with dedicated log search and tools), and even trigger builds. CircleCI's server is squarely aimed at failure triage and config: its headline use is fetching detailed build-failure logs for a broken job so the agent can propose a fix in the same conversation, and it adds finding flaky tests from historical runs, validating config, reading test results and latest pipeline status, running pipelines, rerunning workflows, and rolling back. So the choice tracks the platform you run and whether you want broad pipeline management or sharp failure-and-config tooling. Here is how they compare for an agent.

How they compare

Dimension	Buildkite	CircleCI
Platform	Buildkite Pipelines and Test Engine — clusters and queues for self-managed agents.	CircleCI's cloud CI/CD pipelines and workflows.
Emphasis	Broad pipeline management: list, inspect, create, and update pipelines, read builds, and trigger builds.	Failure triage and config: pull build-failure logs, find flaky tests, validate config, and propose fixes.
Infrastructure visibility	Strong — enumerate organization, clusters, and queues, reflecting Buildkite's self-hosted agent model.	Focused on projects, pipelines, jobs, and workflows rather than agent infrastructure.
Logs and tests	Job logs with dedicated log search tools, plus Test Engine data.	Detailed failure logs and historical flaky-test analysis aimed at fixing broken runs.
Run controls	Trigger builds and manage pipelines as part of the broad surface.	Run pipelines, rerun workflows, and run rollback pipelines for recovery.
Best-fit task	Managing and inspecting Buildkite pipelines and infrastructure, including creating pipelines.	Diagnosing a failing CircleCI job, finding flaky tests, and validating config to ship a fix.

Verdict

Choose by the platform you run and the job you want done. Buildkite's server is the pick when your CI is Buildkite and you want broad management — enumerate clusters and queues, list, inspect, create, and update pipelines, read builds, and trigger them — fitting Buildkite's self-managed agent model. CircleCI's server is the choice when failure triage is the priority: it fetches detailed failure logs, finds flaky tests, validates config, and can rerun or roll back, so the agent can diagnose and propose a fix in one conversation. Both are official; the split is Buildkite's broad pipeline-and-infrastructure management versus CircleCI's sharp failure-and-config focus. Match the server to your CI platform and to whether you manage pipelines or chase failures.

FAQ

Which is best for fixing a broken build?: CircleCI's server is built for that — its headline use is pulling detailed build-failure logs so the agent can propose a fix in the same conversation, plus finding flaky tests and validating config.
Can Buildkite's server create pipelines?: Yes. Beyond listing and inspecting, it can create and update pipelines, read builds, enumerate clusters and queues, pull job logs, and trigger builds.