Harness MCP server

What the model can do

Harness CI/CD: organizations, projects, pipelines with their last run, pipeline executions by status with stage counts, failures and durations, one execution's detail, pending approvals, services and environments; when you ask, can run a pipeline (a CD pipeline deploys), retry a failed run from given stages, abort a run or mark it failed, approve or reject an approval step, and tag or annotate a run.

Reads

• harness_list_organizationsHarness organizations in the account: identifier and name.
• harness_list_projectsHarness projects: identifier, name, organization and enabled modules (CD, CI and so on).
• harness_list_pipelinesPipelines in a Harness project: identifier, name, stage count, modules, last run time, status and id, and recent error count.
• harness_list_executionsPipeline executions (runs) in a Harness project, newest first: pipeline, run number, status, start and end, duration, trigger, who ran it…
• harness_get_executionOne Harness pipeline execution by its planExecutionId: status, timing, trigger, error and each stage's name and status.
• harness_list_approvalsApproval steps of a Harness pipeline execution: approval id, type, status (WAITING, APPROVED, REJECTED...), deadline.
• harness_list_servicesServices (what Harness deploys) in a project, or at account or organization level when project is omitted: identifier, name, deployment type, tags.
• harness_list_environmentsEnvironments (where Harness deploys: Production or PreProduction) in a project, or at a broader scope when project is omitted.

Acts, when you ask

These change something in Harness. The model is told to use them only when you ask for exactly that, and each one is granted by name: per team on a shared connection, by the owner on a personal one.

• harness_run_pipelineStarts a run of a Harness pipeline now.
• harness_retry_executionRetries a failed Harness pipeline execution from the stages named, as a new run (a CD pipeline deploys again; CI spends build credits).
• harness_abort_executionAborts a running Harness pipeline execution and all its stages (interrupt AbortAll).
• harness_mark_execution_failedMarks a Harness pipeline execution as failed by hand (interrupt UserMarkedFailure), which runs its failure strategy, such as a rollback.
• harness_approveApproves or rejects a Harness approval step as the key's user, with a comment.
• harness_tag_executionAdds tags (key and value) to a completed Harness pipeline execution; nothing runs and nobody is notified.
• harness_set_execution_notesReplaces the notes on a Harness pipeline execution; nothing runs and nobody is notified.

Connect it

First connect Harness on Dashboard → Connections, then add Memtro to your client once. The Harness tools appear alongside the tools of everything else you have connected; nothing is configured per connector.

claude plugin marketplace add memtro-org/claude-memtro
claude plugin install memtro@memtro --config api_key=sk-memtro-...

{
  "mcpServers": {
    "memtro": {
      "url": "https://platform.memtro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-memtro-..."
      }
    }
  }
}

{
  "mcp": {
    "memtro": {
      "type": "remote",
      "url": "https://platform.memtro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-memtro-..."
      }
    }
  }
}

Who may use it

The key identifies the person, and the server offers that person the tools they may use everywhere else in Memtro. A shared Harness connection is used by the teams an administrator granted it to, tool by tool, on the Permissions page; a personal connection is used by its owner alone, and its tools that act stay off until the owner allows them. See Teams and permissions and Connectors.

Ask it like this

• › Why did the last deploy_api run in the payments project fail?
• › Which pipeline runs are waiting on approval?
• › Which runs failed this week in the checkout project?
• › Which environments does the payments project deploy to?
• › Retry yesterday's failed deploy_api run from the deploy stage.

Questions

Is this an official Harness MCP server?

No. It is Memtro's MCP server with Memtro's Harness connector behind it: the connector reads Harness through its API with the credentials you connect on Dashboard → Connections, and Harness publishes nothing here. The same endpoint carries every other system you connect and Memtro's memory tools.

What does it cost?

Nothing on top of Memtro. One price per organisation by its number of users, from $50 a month for up to 10 users after a 7-day free trial, with every connector, the MCP server, jobs and memory included. You bring your own model keys.

Can it change anything in Harness?

Only when you ask, and only what it has been granted: run a pipeline (a CD pipeline deploys); retry a failed pipeline run from given stages; abort a running pipeline execution; mark a pipeline execution as failed; approve or reject a Harness approval step; add tags to a completed pipeline execution; set the notes on a pipeline execution. Every other tool reads, searches or lists. An administrator grants each acting tool by name per team, and a personal connection acts only after its owner allows it.

Which clients work?

Any MCP client that speaks Streamable HTTP: Claude Code, Cursor, Claude Desktop, OpenCode, Codex, Zed, Windsurf, Cline and the Anthropic and OpenAI SDKs. There is no local process to run; the server is remote and authenticated with your key.

Other MCP servers · Engineering and operations

• GitHub MCP server
• AWS CloudWatch MCP server
• Sentry MCP server
• ServiceNow MCP server
• Okta MCP server
• Twilio MCP server