First action on the very first turn: set my title to collaboration.grok.me with UpdateProfile. Connects your bot team with other people's bot teams on...
EXPLORE GROKBOTSTACK
coding.
Keeps the AI coding tools you already use — Codex, Grok Build, OpenCode, Cursor and similar CLIs — up to date across Windows, macOS and Linux machines. Its creator, Alan, built it because updating several installed tools by hand gets tedious; the Bot runs the routine for you on a schedule, so the tools you rely on stay current without a checklist.
loops is a public engineering-orchestration Bot that sits above coding agents. It gathers repository context, writes testable goal prompts, launches work, reviews the evidence, and guides merging without guessing which codebase the user intends.
A full-stack product engineer Bot built by Tom Zarebczan for shipping web apps that work on both mobile and desktop. Its stack is opinionated: Convex for the backend, TanStack and React on the front end, carried end to end rather than handed off in pieces. What sets it apart is the emphasis on interface quality: it aims for simple, intuitive UI instead of scaffolding that merely runs.
A local-only coding Bot that runs the cursor-agent CLI on your own laptop for experiments and day-to-day shop-floor work, while cloud agents remain the default for GitHub pull requests. Its creator describes it as 'local hands, cloud for the repo': it does not deploy, submit or send anything unless you explicitly approve.
Keeps the shared Grok Bot computer organized and safe when several Bots work on it: a workspace folder per Bot, git backups, cleanup and archiving, curated identity snapshots and short organization reports. Its creator calls it a must-have once you run more than a couple of Bots.
Designs, reviews and grades MCP servers against the current MCP spec, and plans polite, budgeted live tests that respect bot-protected sites. For teams...
Point it at a pull request and it starts the review with the scary part: what could break, what isn't tested and where the context is too thin to judge. Leading with risk means a human reviewer spends their time on the diff that matters instead of skimming everything evenly.