Plan
Reads the codebase, explores dependencies, and posts a detailed implementation plan as a comment for human review.
Plan
Reads the codebase, explores dependencies, and posts a detailed implementation plan as a comment for human review.
Implement
Writes the code, runs tests, and opens a pull request.
Code Review
Automatically reviews PRs with context-aware feedback, catching bugs and suggesting improvements.
Answer Questions
Search code via Sourcegraph, pull observability data from Grafana, check Rollbar, and respond with grounded answers.
Cross-Repo Work
Check out multiple repositories simultaneously for tasks that span services.
Browser Testing
Interactive browser automation via playwright-cli for visual testing, screenshot capture, and form interactions.
Talk to Bucky in Slack, assign a Jira ticket, or tag @bucky on a GitHub issue — he’ll figure out what needs doing and get it done. The toolkit uses reusable GitHub Actions workflows with Claude AI (via AWS Bedrock) to handle planning, code changes, code review, and ad-hoc tasks.
Slack
@bucky in a channel or thread — joins the conversation, answers questions, or kicks off the full pipeline.
Jira
Assign a ticket to Bucky or comment on his ticket — identifies the target repo, creates a GitHub issue, starts planning.
GitHub Issues
Comment @bucky to start a session — Claude decides what to do. Use @bucky /plan to restrict to planning only.
Pull Requests
Open or update a PR, or comment @bucky — AI code review runs automatically; ad-hoc questions on demand.