Atlas connects to Jenkins by making Jenkins the MCP server itself. Install the mcp-server plugin on your Jenkins controller and the controller exposes build status and console logs at a streamable HTTP endpoint, so Atlas can find the stage that broke and patch the Jenkinsfile. Register it with atlas mcp add jenkins --url https://jenkins.example.com/mcp-server/mcp, authenticate with Jenkins HTTP Basic using a user API token, and turn it off again by setting enabled to false on the mcp.jenkins entry in atlas.json.
What the Atlas and Jenkins integration is
Jenkins becomes the MCP server itself once you install the mcp-server plugin, exposing build status and console logs on the controller. In 2026 that means Atlas talks straight to your Jenkins controller, finds the stage that broke, and patches the Jenkinsfile without any middleware process in between.
Most integrations put a separate broker between the agent and the system. Jenkins does not. The mcp-server plugin runs inside the controller you already operate, so the thing serving build status and console logs is the same JVM that ran the build. For Atlas that shortens the trust chain: no extra container to keep alive, no third party holding your Jenkins credential. For you it means the integration lives where your Jenkins administration already lives, on the controller, behind whatever network boundary already protects it.
Installing the mcp-server plugin and registering the controller
Install the mcp-server plugin on your Jenkins controller, then add it with atlas mcp add jenkins --url https://jenkins.example.com/mcp-server/mcp. Those 2 steps run in that order, because until the plugin is present the controller serves no MCP endpoint and Atlas has nothing to connect to.
Substitute your own controller hostname for jenkins.example.com. The path is what matters: /mcp-server/mcp is the endpoint the plugin publishes on the controller. If your Jenkins sits behind a reverse proxy, confirm the proxy forwards that path and does not strip it, because a 404 at registration time almost always traces to a proxy rule rather than to the plugin. Once atlas mcp add jenkins succeeds, Atlas lists the Jenkins tools alongside its other servers and can query build status directly.
Authenticating Atlas against Jenkins with an API token
Auth is Jenkins HTTP Basic with a user API token, so pass --header 'Authorization=Basic BASE64_OF_USER_COLON_APITOKEN' when you register the server. Jenkins does not use OAuth here, and it does not accept your login password: the API token from your Jenkins user account is the credential.
The header value is the base64 encoding of your Jenkins username, a colon, and the API token, which is standard HTTP Basic. Generate the token from your Jenkins user configuration page and treat it as a secret. The account behind the token defines what Atlas can read, so a service account scoped to the jobs you actually debug is a better choice than a Jenkins administrator account. If the connection returns 401, regenerate the token and re-encode the header rather than guessing at the plugin configuration.
Choosing the right Jenkins MCP endpoint
The Jenkins controller exposes 3 MCP endpoints, /mcp-server/mcp, /mcp-server/sse, and /mcp-server/stateless, and the streamable HTTP endpoint at /mcp-server/mcp is the recommended one. Point Atlas there first, and reach for the other two only when a proxy or runtime constraint makes it unworkable.
Three endpoints exist because MCP transports evolved, and the plugin keeps the older shapes available. /mcp-server/sse is the server-sent-events transport. /mcp-server/stateless drops session state. /mcp-server/mcp is the streamable HTTP endpoint and the one to use by default with Atlas. Picking the recommended endpoint avoids a class of confusing failures where the connection opens but tool calls hang, which usually means the transport in the URL and the transport the client expects do not agree.
The daily loop: from a red Jenkins build to a patched Jenkinsfile
Ask Atlas to read the failing build's console log and name the stage that broke. Then let Atlas patch the Jenkinsfile and show the diff before you commit it. Those two requests are the whole 2026 loop, and the console log grounds both of them in what Jenkins actually printed.
Naming the stage first is what keeps the fix honest. A Jenkins console log is long and noisy, and an agent that skips straight to a patch will pattern match on the loudest error rather than the causal one. Force the intermediate answer: which stage failed, and on what line of output. Once Atlas has named it, ask for the Jenkinsfile patch and read the diff in the terminal. Commit only after you have compared the diff to the quoted console output.
Turning the Jenkins MCP server off quickly
Turn the Jenkins server off quickly by setting enabled to false on the mcp.jenkins entry in atlas.json. That 1 edit revokes Atlas's access to build status and console logs without uninstalling the mcp-server plugin from the controller or regenerating the Jenkins user API token.
A kill switch that lives in your own config file, rather than in Jenkins administration, matters during incidents. If you need Atlas to stop touching CI while a release is in flight, flip enabled to false on the mcp.jenkins entry and Atlas simply stops seeing those tools. The plugin keeps running for anyone else who uses it, the token stays valid, and re-enabling is a one-character edit. Keep the token rotation option in reserve for an actual credential compromise.
Setup
- 01Install the mcp-server plugin on your Jenkins controller, then add it with atlas mcp add jenkins --url https://jenkins.example.com/mcp-server/mcp
- 02Authenticate with Jenkins HTTP Basic using a user API token by passing --header 'Authorization=Basic BASE64_OF_USER_COLON_APITOKEN'
- 03Prefer the streamable HTTP endpoint at /mcp-server/mcp, since the controller also exposes /mcp-server/sse and /mcp-server/stateless but that one is recommended.
- 04Ask Atlas to read the failing build's console log and name the stage that broke.
- 05Let Atlas patch the Jenkinsfile and show the diff before you commit it.
- 06Keep the kill switch handy: set enabled to false on the mcp.jenkins entry in atlas.json to turn the server off quickly.
Frequently asked questions
- is there a Jenkins MCP server
- Yes. Jenkins becomes the MCP server itself once you install the mcp-server plugin, which exposes build status and console logs on the controller so Atlas can find the stage that broke.
- how do I connect Atlas to Jenkins
- Install the mcp-server plugin on your Jenkins controller, then run atlas mcp add jenkins --url https://jenkins.example.com/mcp-server/mcp with your own controller hostname.
- how do I authenticate the Jenkins MCP server
- Auth is Jenkins HTTP Basic with a user API token, so pass --header 'Authorization=Basic BASE64_OF_USER_COLON_APITOKEN'. Jenkins does not accept your login password for this.
- which Jenkins MCP endpoint should I use
- The streamable HTTP endpoint at /mcp-server/mcp is the recommended one. The controller also exposes /mcp-server/sse and /mcp-server/stateless, but use those only when a proxy or runtime constraint forces it.
- can an AI agent fix a failing Jenkins build
- Ask Atlas to read the failing build's console log and name the stage that broke, then let Atlas patch the Jenkinsfile and show the diff before you commit it.
- how do I disable the Jenkins MCP server in Atlas
- Set enabled to false on the mcp.jenkins entry in atlas.json. That revokes access immediately without uninstalling the mcp-server plugin or rotating the Jenkins user API token.
- does Atlas need a separate server to read Jenkins logs
- No. The mcp-server plugin makes the Jenkins controller itself the MCP server, so there is no extra broker process between Atlas and your build status or console logs.
Try Atlas in your terminal
The terminal-native AI coding agent. Free core, single binary.
Install AtlasRelated guides
Atlas for Laravel in 2026
Atlas is a terminal-native AI coding agent for Laravel in 2026. Run it in a Laravel app with an artisan file, add a controller or Pest test, and review the diff.
Atlas vs Cursor: terminal AI coding agents compared (2026)
A grounded 2026 comparison of Atlas and Cursor across workflow, change review, extensibility, and pricing for developers choosing an AI coding agent.
Atlas with CodeGemma 7B (Ollama): Fill-in-the-Middle on 8GB in 2026
CodeGemma 7B (Ollama) is Google's 5.0GB code model with fill-in-the-middle training and an 8K context, Free (self-hosted). Atlas setup and honest tradeoffs for 2026.
Atlas with GLM-5.1: Reasoning, Cost, and Context in 2026
GLM-5.1 from Z.ai drives Atlas with a 200,000 token context at $1.40 per Mtok input and $4.40 per Mtok output. Setup, honest tradeoffs, and how it compares to GLM-5.2.
Atlas vs Fine.dev: Terminal AI Coding Agents in 2026
Compare Atlas and Fine.dev for terminal AI coding in 2026. Atlas offers a free core TUI with permission-gated tools, while Fine.dev provides asynchronous cloud agents and prebuilt libraries.
Atlas vs v0: A Developer's Guide for 2026
Comparing Atlas and v0 in 2026: Atlas offers terminal-native AI coding with explicit diffs and BYO model keys, while v0 provides a visual editor for React/Next.js apps.
Atlas with GLM-5.2: A 1M Token Open-Weights Model at $1.40 per Mtok (2026)
GLM-5.2 drives Atlas on a 1M token context at $1.40 / $4.40 per Mtok. Z.ai's June 2026 flagship, the first GLM to reach 1M, with open-weights lineage.
Run the Test Suite and Triage the Failures with Atlas in 2026
How to triage a failing test suite with Atlas in 2026: bash truncates at 2000 lines or 50 KB and saves the full log, then grep groups failures by root cause.