MCP

Automate launch filing

Set up editor or CI automation so every branch files a Buron source without anyone running /launch manually.

Running /launch manually works well for individual features. But if you want sources filed automatically, there are two paths: editor automation and CI.

Path 1: editor automation

The fastest way to automate. Your editor runs /launch on a recurring basis while you work, with no CI config and no API keys in GitHub.

Claude Code supports routines: recurring tasks that run on a schedule. Set one up to run /launch on the current branch whenever you're working.

Cursor has background agents that can run skills automatically on a schedule.

In both cases, the /automate-launch skill walks you through the setup:

/automate-launch

Once configured, sources get filed as you work. No extra steps, nothing to commit.

Path 2: CI automation

For teams that want coverage on every PR regardless of editor, a GitHub Actions workflow runs /launch through an AI agent.

Scaffold the workflow

From your project folder, run:

buron setup-ci

The CLI asks which AI agent to use for CI:

  • Claude Code: uses anthropics/claude-code-action@v1
  • Cursor: runs a launch script via the Cursor SDK
  • OpenAI Codex: runs a launch script via the OpenAI Agents SDK

It creates .github/workflows/buron.yml and sets BURON_TOKEN as a GitHub secret.

Add your AI provider key

The CLI prompts you to add your provider's API key as a GitHub secret. If you have the gh CLI installed:

gh secret set ANTHROPIC_API_KEY

This prompts securely for the value without exposing it in shell history.

If you don't have the gh CLI, install it with brew install gh (macOS) or winget install GitHub.cli (Windows). Or skip it: the setup command prints a link to your repo's secrets settings page for manual entry.

How it works

On every pull_request event, the workflow runs in your CI environment using your AI provider key:

  1. Checks out the repo
  2. Installs the Buron CLI
  3. Your agent reads the PR diff, comments, and code context, then runs /launch
  4. The CLI pushes the source to Buron

Buron never sees your code. The agent runs entirely in your GitHub Actions runner, and only the final source, a plain-language summary of the change, is sent to Buron.

Sources accumulate passively. The team ships PRs as normal, and launches appear in the dashboard without anyone thinking about it.

Which path to pick

Editor automationCI automation
SetupRun /automate-launchRun buron setup-ci, add API key
Covers every PROnly when the editor is openYes, automatically
Needs AI provider API key in CINoYes
Runs without developer actionYes, while the editor is runningYes
Best forMost teams, solo developersLarge teams, full PR coverage

Editor automation is the simplest starting point. It works out of the box with no CI changes. Add CI later if you need guaranteed coverage on every PR.

Next steps

Was this page helpful?

On this page