Run a multi-bot team that opens draft PRs
A Grok setup of research, classifier, builder, tester, reviewer, and security bots maps repos, makes minimal changes, runs tests and lint, and opens draft PRs; the author reports over 1,000 PRs.
Done withGrok Bot
- What they did
- The author set up six sequential Grok bots. The first reads the issue and maps the repo to relevant files and prior PRs, passing only useful context. The second classifies the task and risk level to route it. The builder writes the smallest possible code change. A tester bot then runs tests, lint, type checks and builds, feeding failures back to the builder. A separate reviewer examines the diff for regressions and missing tests. A final security bot checks secrets, auth and dependencies before opening a draft PR.
- How it went
- The loop produced over 1,000 PRs. No shortfalls are stated.
Try it yourself with Grok Bot
Set up a team of bots for my repo [repo URL]: one maps the codebase, one classifies the task from [issue tracker or label], one builds a minimal change, one runs tests and lint, one reviews, and one checks security. Each task should end in a draft pull request with a summary of what changed and the test results. Never merge or push to the main branch; leave every PR as a draft for me to review.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
Source: X · Oct 4, 2026
Related use cases
Plan and book a short-notice work trip
Grok Bot researched flights and booked the one Vince picked, though he had to take over the computer to fill in data and found it harder than booking himself.
Run content creation with specialized bots
Creators use bots to study outlier YouTube videos, and run newsletters with research, drafting, editing and scheduling bots; one experiment made 44 thumbnail proposals.
Create a subtitled video from one prompt
Agent Mode in Grok Imagine made a 36-second video with matching scenes from different angles, a storyline and subtitles.
Build a C compiler with a team of parallel agents
Anthropic's engineering write-up on tasking Opus 4.6 agent teams with building a C compiler.
Hand narrow code changes to Devin for pull requests
A team used Devin for a month on small fixes like chart formatting and API edge cases, with tests and PRs; a complex full-stack task went poorly.
Build a booking and payment app for tattoo artists
A former programmer used Replit Agent over a weekend to build a booking app with calendar, Google OAuth, Stripe deposits and notifications, though time zones and UI needed back-and-forth.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.