Gleap AI agents can use CircleCI through Composio, for engineering teams that run CI on CircleCI and fix bugs that customers report in Gleap. Kai Code and custom agents can read pipelines, failed jobs and test results, so everyone can see why the checks on a fix passed or failed.
What a Gleap agent can do in CircleCI
- Why a check failed: Get Job Details and Get Test Metadata show which step and which tests failed in a pipeline, for example on a pull request Kai Code opened.
- Flaky or real: Get Flaky Tests lists the tests CircleCI marks as flaky, so a failing check can be told apart from a regression.
- Where a fix stands: List Pipelines for Project and Get Workflow Summary show whether the pipeline for a fix’s branch is running, passed or failed.
- Test reports and screenshots: Get Job Artifacts returns the files a job stored, such as test reports or browser screenshots.
Example: a red check on a fix for a customer
Kai Code opens a pull request for a date picker bug a customer reported in Gleap, and a CircleCI check fails. The engineer asks Kai Code to look at it, and with CircleCI in Kai Code’s MCP settings it reads the failed job and test metadata. The failure is in a test CircleCI lists as flaky, so Kai Code reports that and leaves the code unchanged. The engineer reruns the check and reviews the pull request.
How to connect CircleCI
- In Gleap, open Kai Code, click Settings and open MCP.
- Click Add MCP server, search for “CircleCI” and select it.
- Enter a CircleCI personal API token and any other details the form asks for, then click Connect CircleCI.
- Open the CircleCI server and switch off the tools Kai Code should not call, such as Trigger Pipeline, the environment variable and context actions and Delete Project.
- For a custom agent, add CircleCI under the agent’s Tools with Add tool instead, then publish and test it.
Kai Code is on Team, Pro and Enterprise, and Kai Support can use connected apps as tools on the same plans. Custom agents are on the Team, Pro and Enterprise plans.
FAQ
Does Gleap integrate with CircleCI?
Yes. Kai Code and custom agents use CircleCI through Composio’s CircleCI toolkit, authenticated with a CircleCI API token. They read or trigger CircleCI only while they run; Gleap does not receive CircleCI webhooks.
What can a Gleap AI agent do in CircleCI?
It can list and read pipelines, workflows, jobs, test metadata, flaky tests, artifacts, insights and project settings, trigger pipelines, and manage contexts, environment variables, orbs and projects. Environment variables and contexts hold secrets, so keep those tools off.
Which Gleap plan includes the CircleCI integration?
Kai Code can use CircleCI on Team, Pro and Enterprise, and Kai Support can use connected apps as tools in its customer conversations on the same plans. Custom agents that use CircleCI are part of the Team, Pro and Enterprise plans. The token belongs to a CircleCI user, and the agents see what that user can see. AI usage is billed separately; see pricing.
Does Kai Code need CircleCI to open pull requests?
No. Kai Code works in your repository and opens pull requests without it. Connecting CircleCI lets it read your CI results when someone asks it to look at a failing check.
