Skip to main content

​
Overview

Code quality and CI/CD automations help maintain high standards and reduce manual intervention when builds fail. These automations can detect issues, attempt fixes, and keep your development workflow smooth.

​
Example: CI Failure Auto-Fix Monitor

Use Case: Automatically detect and attempt to fix failing CI checks on pull requests. How it works:
  • Monitors pull requests for failing CI checks
  • Analyzes failure logs to understand the issue
  • Attempts to fix common issues (linting, formatting, type errors)
  • Pushes fixes to the PR branch
  • Comments on the PR with what was fixed
Trigger: PR check failure webhook Prompt Template: 
A CI check has failed on PR #{pr_number}.

1. Fetch the CI logs for the failing check
2. Analyze the error messages to determine the type of failure:
   - Linting errors
   - Type errors
   - Test failures
   - Build errors

3. If it's an auto-fixable issue (linting, formatting):
   - Clone the PR branch
   - Run the appropriate fix command (pnpm run lint --fix, pnpm run pretty)
   - Commit and push the fixes
   - Comment on the PR: "🤖 Automatically fixed {issue_type}"

4. If it's not auto-fixable:
   - Comment on the PR with a summary of the error
   - Tag the PR author
   - Suggest potential solutions based on the error

5. Update the PR status comment to show Tembo is working on it
MCP Servers Needed: GitHub, Linear Safety Considerations:
  • Only auto-fix low-risk issues (formatting, linting)
  • Always commit fixes separately with clear messages
  • Never auto-fix test failures or build errors without human review

​
Tips for Code Quality & CI/CD Automations

​
Start with Low-Risk Fixes

Begin with automations that only fix formatting and linting issues. These are safe to auto-fix and won’t break functionality.

​
Always Notify the Team

When an automation makes changes to code, always leave a comment explaining what was fixed and why. This maintains transparency.

​
Set Clear Boundaries

Define explicitly what the automation can and cannot fix. For example, never auto-fix failing tests without human review.

​
Track Success Rates

Monitor how often your automation successfully fixes issues versus how often manual intervention is needed. Use this data to improve your prompts.

​
Integrate with Issue Tracking

For issues that can’t be auto-fixed, automatically create tickets in Linear or Jira with details about the failure.