THIS IS ACTUAL F*CKING GOLD Grok Bot's own docs describe Bots handing work to each other. Here's the coding setup I'd build around that. The idea: run Builder, Scout and Reviewer roles together, each one doing only the job it's best at. Reviewer-level judgment on every big call, without pulling the Reviewer into every routine turn. You just start the Grok Bot team like this: Builder own the code. Scouts handle lookup. Reviewer stay on call for the big decisions. here's the split: → Builder runs the coding task and writes the changes → Scout Bots explore files, call sites and the draft diff in parallel → Reviewer stays on call as advisor for the big calls → you stop flooding the lead thread with every lookup transcript Reviewer only steps in at three moments: → before a plan locks → when the same error shows up twice → before calling it done Builder builds. Scouts explore. Reviewer reviews. Tool routing is the same move one layer down: the forks that need no new judgment follow explicit rules, and the Bots see the ones that need a decision. the full tree > Builder owns the coding task and edits the source > search maps the repo as a Scout Bot > grep finds call sites as a Scout Bot > audit checks the proposed diff in a scratch copy as a Scout Bot > Reviewer stays on call and reads the shared handoff when consulted > explicit routing rules cover file choice, tool choice and when to ask paste the tree and this prompt into Grok Bot ↓ "Rebuild my Grok Bot coding setup around this tree: 1. Give Builder ownership of implementation and Reviewer ownership of strategic review. 2. Propose three Scout Bots: search maps the repo, grep finds call sites, and audit checks the proposed diff. Assign each a clear output and return concise findings with file references. 3. Add one operating rule: consult Reviewer before locking a plan, after the same error twice, and before calling a task done. Include the plan, latest diff, evidence and unresolved questions in the handoff. 4. Keep mechanical routing in explicit rules. Use a separate scratch copy for audit checks and return findings to Builder. Show the proposed roster and every configuration or file change as a diff first. No edits or Bot creation until I say go." ↳ docs.x.ai/grok-bot/chat-and-… save it, you'll use this every single day.

Gemini vs Opus 5.5 vs ChatGPT 6 vs Grok same brief: a Trojan horse in Blender > Gemini $4.87 > Opus $4.94 > ChatGPT $4.96 > Grok $4.99 all look decent, but imo Grok wins on detail and atmosphere full prompt in the reply ↓

A Chinese fantasy RPG by Kimi K3.1 Pro Took 8 hours and 3.7M tokens Total cost: $33.68 Ancient temples, floating mountains and a river guarded by giant statues Astra / Opus 5.5 could have easily cost over $450 And this is what it looked like when Kimi added cinematic fog and water reflections

HOLY SH*T, I CHANGED ONE FIELD NAME AND MY GROK BOTS RIPPED A $96,000 WORKFLOW IN HALF i thought i'd cleaned up a spreadsheet. invoice_id became invoice_ref. looked nicer. went to bed. 64 receipts stopped matching their invoices. $1,500 each. $96,000 sitting in suspension while i slept very professionally. Finance couldn't release them. Builder wanted to guess the mapping. Guard made it show the originals first. the replies said the dashboard was just pretty wires. then one wire broke and every downstream task had to wait. the repair: → isolate the broken invoice join → keep all 64 original payment events → propose a mapping repair and replay the queue → i approve the new mapping → Review matches 64 out of 64 → $96,000 released from suspension that money had already been received. fixing my spreadsheet didn't create another $96,000, although i briefly considered taking credit for it. at 0:08 the fabric starts ripping down the middle. at 0:12 the broken join is isolated. at 0:16 the agents start stitching a gold seam across the gap. at 0:22 everything matches again. at 0:23 a sock falls out of the repaired workflow. finally an explanation for the washing machine. i wanted agents that could fix my work. apparently i also needed agents that could prove i was the problem. the repair loop and approval step are below. send this to whoever says “i only renamed one column”.

YOUR GROK BOT TEAM CAN FINISH EVERY TASK AND STILL TRASH THE FINAL FILE Two writers start from the same version. One adds the outline. The other adds examples. Both aim at plan.md. Now ask which version survives the handoff. Watch 0:12. In this illustrated workflow, the purple Reviewer catches both drafts targeting the same destination. Everything pauses before either draft becomes the final file. I'd give the team four explicit jobs: → Writer A owns draft-a.md → Writer B owns draft-b.md → Chief is the only owner of the final plan.md → Reviewer compares that file against BOTH drafts At 0:18, the writers have separate handoffs. Chief combines the sections. The final check asks whether the outline AND the examples actually made it through. Grok Bot's account computer and files are shared across Bots. Giving two Bots different names doesn't assign ownership of a shared document. That ownership rule belongs in the setup, before the work starts. Keep the original. Keep both drafts. Make one Bot responsible for assembling the final file. “All agents done” can look spectacular on a dashboard while telling you very little about the deliverable. Steal the proposed handoff rules in the first reply.

YOUR GROK BOT IS ONE OUTDATED SOURCE AWAY FROM SHIPPING VERY CONVINCING BULLSHIT Four Bots. A dashboard full of activity. A report coming together. Then the purple Bot stops the handoff over ONE missing word. Watch 0:12. In this illustrated Grok Bot workflow, a saved note says a feature is available. The current page still calls it beta. That qualification is about to disappear into a confident-looking brief. The Checker holds the claim. Reader goes back to the current page. Editor restores the qualification and attaches the source. The team has four jobs: → Reader collects links and dated excerpts → Editor builds the brief → Checker challenges changing claims → Chief delivers the checked draft with unresolved questions Give the Checker permission to interrupt the whole process over one unsupported sentence. A source link can look reassuring even when nobody has reopened it. That’s the part I’d worry about before adding another hundred agents. Save this before your next AI-generated report.

STOP MAKING ONE GROK BOT DO EVERY DAMN JOB Finding the right files, writing a patch and questioning the diagnosis are different jobs. Give each one an owner. Grok Bot lets named Bots coordinate through messages and shared files. Here's a small coding setup I'd use: Builder, Scout and Reviewer. Create the three Bots, put them in a group, then tell Builder: "Own this fix. Ask Scout for file references. Request Reviewer input before locking the plan, after the same error twice, and before handing me the final diff." Keep the responsibilities clear: → Builder owns the source changes and runs the checks → Scout maps the repo, finds call sites and reads the relevant docs → Reviewer challenges the plan and checks the evidence → each handoff names the files and the next owner The second failed patch is where this gets useful. That's a good moment for another Bot to question the cause before Builder tries again. Watch 0:12 in the illustrated run: the same test has failed twice. Builder requests a review, receives a diagnosis to investigate, then revises the patch and reruns the checks. The setup to copy: > Builder owns the proposed diff > Scout leaves file references in a findings file > Reviewer receives the task, evidence and current diff > Builder applies the agreed revision > the final handoff includes check results and open questions Paste this into Grok Bot: "Help me set up Builder, Scout and Reviewer for a coding task: 1. Give each Bot the responsibility above. Keep project notes under /workspace/project and use one owner for source changes: Builder. 2. Instruct Scout to inspect files and docs without changing source. Have it return exact references and write findings.md. 3. Ask Builder to request Reviewer input before committing to a plan, after the same error twice, and before final handoff. Include the task, findings and proposed diff in the request. 4. Have Builder run relevant checks and return the diff with evidence. These are working instructions, not guaranteed automatic gates. Show me the proposed setup first. Wait for approval before creating Bots or changing project files." All Bots on your account share the cloud computer. Read-only Scout behavior here is an instruction, not a security boundary. docs.x.ai/grok-bot/bots docs.x.ai/grok-bot/chat-and-… Save this for the next fix that keeps coming back.