stable URL: /l/chai-bot

Chai Bot — AI assistant for kernel developers

How to use Chai Bot for CKI pipeline questions, Jira navigation, code exploration, and kernel workflow help

Chai Bot is an AI assistant available in Slack and via MCP (Model Context Protocol). It helps kernel developers navigate CKI pipelines, query test results, manage Jira issues, and explore cki-project repositories without leaving their usual workflow.

Where to find Chai Bot

Chai Bot is present in several Slack channels:

  • #team-kernel-cki — CKI team discussions and pipeline questions
  • #help-beaker — Beaker and test infrastructure help
  • #team-kernel-se — Sustaining Engineering kernel topics

You can also send Chai Bot a direct message in Slack.

How to use Chai Bot

Mention @chai-bot in any Slack thread to ask a question. Chai Bot replies in the thread so the conversation stays organized. You can also start a direct message conversation for private questions.

Before Chai Bot processes your messages, you must grant consent by typing /consent in a direct message to the bot. This is a one-time step required by policy.

What Chai Bot helps with

Chai Bot assists kernel developers across several areas:

  • CKI pipeline questions and debugging — understand pipeline stages, find why a build or test failed, and navigate pipeline configurations.
  • DataWarehouse results navigation — query and explore test results stored in the CKI DataWarehouse.
  • Jira issue management — search, read, and update issues in the CKI, KMAINT, and RHELTEST projects.
  • Code navigation — browse and search across cki-project repositories to understand tooling and pipeline code.
  • Beaker and Testing Farm questions — get help with test infrastructure, job submission, and result interpretation.
  • Kernel workflow and MR label automation — understand merge request labels, subsystem ownership, and CI automation for kernel trees.

MCP access for IDE integration

Chai Bot is available as an MCP server, which lets you connect it to your local development environment. Supported clients include Codex CLI, Claude Code, and Cursor.

To get started:

  1. Open the Home tab of the Chai Bot app in Slack.
  2. Generate a self-service MCP token from the Home tab.
  3. Configure your IDE client to connect using the token.

This gives you the same capabilities as the Slack interface directly inside your editor.

Teaching Chai Bot with the learn command

You can contribute domain knowledge to Chai Bot using the learn: command. When you include learn: at the start of a message, Chai Bot stores the information for future reference. This is useful for recording team conventions, troubleshooting steps, or process details that are not yet documented elsewhere.

Getting help from the CKI team

If Chai Bot cannot answer your question or you need human assistance, file an issue in the CKI Jira project.