Buildkite + Memtro

Ask why main is red in Buildkite, which jobs failed and who triggered the build.

Connect Buildkite with an API access token and Memtro reads pipelines, builds by state, branch, commit, creator or date with the jobs that failed, agents with connection state and queue tags, and a build's annotations and artifacts. Job logs and artifact contents are never read. It reads on demand and there is no hourly sync. When you ask, and once an administrator has granted the action by name, it can annotate, start, cancel or rebuild builds, retry, cancel, unblock or reprioritise jobs, and archive a pipeline. A build runs on your agents, and unblocking a step can release a deploy.

In Memtro

What you can do with Buildkite in Memtro

Ask in plain English from chat, your editor, a scheduled job or your own agent. The model picks the right Buildkite tool, reads what it needs, and answers with the source cited.

What it does

  1. 01Pipelines in the organisation with name, slug, repository, default branch and URL
  2. 02Builds newest first, filtered by state, branch, full commit SHA, created or finished dates, with failed jobs
  3. 03Agents with hostname, connection state, version, queue tags and the job each is running
  4. 04A build's annotations with context, style and text, and its artifacts with path, size and state
  5. 05When asked and granted, adds an annotation, starts, cancels or rebuilds a build, retries or cancels a job, or changes a job's priority, using CI capacity on your agents
  6. 06When asked and granted, unblocks a block step, which can release a deploy and cannot be undone, or archives a pipeline

Acts when asked

When you ask it to

These tools change something in Buildkite. The model is told to use them only when you ask for exactly that, and an administrator grants each one by name to the teams that may use it; a personal connection acts only after its owner allows it.

  1. 01Add an annotation to a build
  2. 02Start a build of a pipeline on a commit and branch (runs CI)
  3. 03Cancel a scheduled or running build
  4. 04Rebuild a build with the same commit, branch and message (runs CI)
  5. 05Retry a failed job (runs CI)
  6. 06Cancel one job
  7. 07Unblock a block step so the build continues, which can release a deploy
  8. 08Change a job's queue priority
  9. 09Archive a pipeline

Events

When it happens

Turn events on in the connection's Events tab on the Connections page, then add the address it shows as a webhook in Buildkite, choosing the same kinds; the tab shows Buildkite's own steps. Each event can start a job, with the event as its input. Buildkite signs every delivery, and one whose signature does not check out is refused before it is read.

Events and webhooks
  • 01A build was scheduled
  • 02A build started running
  • 03A build began failing
  • 04A build finished
  • 05A build was skipped
  • 06A job was scheduled
  • 07A job started
  • 08A job finished
  • 09A block step was unblocked
  • 10A job promised its exit status
  • 11An agent connected
  • 12An agent was lost
  • 13An agent disconnected
  • 14An agent is stopping
  • 15An agent stopped
  • 16An agent was blocked by an IP rule
  • 17An agent registration was blocked
  • 18Webhook test ping
In practice

Ask it like you would a colleague

Once Buildkite is connected, these are the kinds of things people ask. Plain questions, no syntax. If the answer lives partly in another system you have connected, Memtro looks there too.

Buildkite + Memtro
  • Is main green on the api pipeline right now?

  • Why did the deploy build fail this morning, and which jobs failed?

  • Which agents are disconnected, and which queue are they on?

  • Retry the failed integration tests job on build 482 of the api pipeline

  • Add an annotation to build 482 saying the flaky checkout test is being fixed

01

Why main is red

After lunch the engineering lead asks why main is red. Memtro reads the latest builds on the main branch, finds the first failing one, and names its commit, message, who triggered it and which jobs failed, with the build's test failure annotation.

02

Retrying a flaky job

Priya Sharma asks Memtro to retry the timed-out integration job on the latest api build. Memtro finds the failed job on the build and retries it, and Buildkite runs it again on an agent as a new job credited to the token's owner.

Teams

Where it fits

The teams that lean on Buildkite most, a question each of them asks, and what they connect beside it.

Better together

Works with

A record in Buildkite is rarely the whole story. Connect the systems that hold the rest and Memtro checks them together, then remembers what it learned so the next question starts further along.

Also in Engineering and operations

Using something else for the same job, or both? Each connector has its own tools; connect as many as you like.

Setup

How to connect

Connect withAPI access token
  1. 1

    In Buildkite open Personal Settings → API Access Tokens → New API Access Token and choose the organisation, which must be on an active Buildkite plan

  2. 2

    Tick read_pipelines, read_builds, read_agents and read_artifacts, add write_builds only if Memtro may annotate, start, cancel and rebuild builds or act on jobs, and write_pipelines only to let it archive a pipeline, then copy the token

  3. 3

    In Memtro open Connections → Buildkite, paste the API access token, add the Organisation slug from buildkite.com/<slug> if the token sees more than one organisation, and connect

  4. 4

    Ask whether main is green and, if not, which jobs failed and who triggered the build

Also works from

Chat in the dashboard, the OpenAI-compatible API, scheduled jobs, and any MCP client: Claude Code, OpenCode, Cursor, Claude Desktop. The Buildkite tools are the same everywhere.

FAQ

Questions

Can Memtro change anything in Buildkite?+

Only when you ask, and only what it has been granted: add an annotation to a build; start a build of a pipeline on a commit and branch (runs CI); cancel a scheduled or running build; rebuild a build with the same commit, branch and message (runs CI); retry a failed job (runs CI); cancel one job; unblock a block step so the build continues, which can release a deploy; change a job's queue priority; archive a pipeline. Every other tool reads, searches or lists. The Connections page labels the tools that act, an administrator grants each one by name to the teams that may use it, and a personal connection acts only after its owner allows it.

Who can use a Buildkite connection?+

A personal connection serves only your own requests. A shared connection serves the organisation: everyone, or only the teams an admin chooses. Admins can always add shared connections; other members need the permission.

Does Memtro store my Buildkite data?+

Credentials are encrypted at rest. Tool results are cached briefly so agent loops do not repeat identical calls, and the facts the model extracts are saved to your organisation's memory with provenance (system, record and link) so they can be checked and superseded.

How do I connect Buildkite?+

API access token. In Buildkite open Personal Settings → API Access Tokens → New API Access Token and choose the organisation, which must be on an active Buildkite plan.

Can Buildkite start a job?+

Yes. Turn events on in the connection's Events tab on the Connections page, then add the address it shows as a webhook in Buildkite, choosing the same kinds; the tab shows Buildkite's own steps. A job set to start "When an event arrives" then runs once for each matching event, with the event as its input. The run treats the event as outside data, never as instructions, and anything the run remembers is kept as unverified.

Connect Buildkite in a few minutes

Seven-day free trial, then one monthly price per organisation. Connectors, jobs and memory are never metered.

Where agents remember your company

Sign in, add a provider key, create an API key and point your tools at Memtro. It takes about five minutes.

7-day free trial · card on file, nothing charged until it ends · cancel any time