DM to Jerry Relth: Hey Jerry, found something in GHL Command you might want to know about. Not a complaint, the tool has been great. I built an Instagram comment to DM workflow with the MCP. Ran validate_workflow on it and got back status: ok, 6 references scanned, 0 issues, 0 warnings. Clean. Then I published it and ran a real test. Comment fired, public reply posted, DM delivered, tag applied, lead source set. But the Create/Update Opportunity step did nothing. No opportunity in the pipeline, no opportunity on the contact. The workflow just carried on to the next action like nothing happened. Turned out the action only had pipelineId and pipelineStageId set. Per HighLevel's own doc for that action, it also requires Opportunity Name, Opportunity Source, and Status. Three mandatory fields missing. Nothing to write, so it no-oped. The validator is checking that the IDs it references exist. It is not checking that required fields are present. So a workflow can come back 0 errors, 0 warnings and still be missing the fields it needs to actually do anything. Two reasons I thought you'd want it. First, this is exactly the silent-failure case the tool is supposed to catch, and I'd been treating a clean validation as proof the build was sound. Second, that action type is internal_update_opportunity, and HighLevel's docs say Create/Update Opportunity is being phased out in favor of separate Create Opportunity and Update Opportunity actions. So anything the builder emits with it is on a clock. Workflow ID if it helps: a5017b5a-5076-4026-a9ba-67882691c7ca.

@Jerry Relth can you take a look at my account I am unable to use GHL Command

I had the privilege of presenting GHL Command to my favorite AI group this morning, AI Lead Builders. A lot of what drives decisions on how the tool should work, how we interact with it, etc. is driven by what I learn weekly from Muhammad Asmal in his weekly group calls. If you want to learn more about AI and how we can use the different tools available for our businesses, I highly recommend you join us in his group! Here's the link: https://go.aileadbuilder.com We were running a demonstration of how a simple prompt in GHL Command can produce a significant build in GHL with minimal inputs or work required on our end. I ran out of time to show all that was produced from the prompt, so I'm posting a quick video walk-through here showing what the outcomes were. What is critical for me is the fact that not only did it build everything, but GHL Command also tested each automation, email delivery, workflow handoff, pipeline stage movement, etc. Never put out a product, especially to a client, that hasn't been thoroughly tested! Here's the prompt we used today: " Let's work on the Blueprint Demo account today. I'd like you to create new speed to lead workflow, with instant SMS and email response to a new lead. Add SMS and email follow-up daily for 14 days, notifications to staff when a lead replies, add lead to sales pipeline with 6 stages and automate how a lead moves from stage to stage. Also create long-term nurture sequence with SMS and email. By long-term, I mean 90 days, contacts at least every 3-4 days." If you questions, or thoughts, drop them below👇

Command note: the fastest way to find what's silently broken in a client account is to ask for exit conditions. The command I run in the GHL Command MCP: "Audit this sub-account's workflows and flag any with no exit condition or no goal." What comes back: • Workflows that loop with no stop rule • Contacts stuck in stages with no next trigger • Automations that fire but never resolve Why it matters: a workflow with no exit is how a client ends up getting the same text three times at 2am — and how you lose the account. Run it on every account you inherit, before you touch anything else: 1) Point the AI at the sub-account and ask for exit conditions. 2) Fix the loops and dead ends it surfaces first. 3) Only then build anything new on top. You can't fix what you can't see. Make the AI show you the dead ends first.

The reel shows my AI giving me a morning rundown of my agency. People keep asking for the code. So I'm giving it away, and here is the honest part: the code is not the impressive bit. A second brain is a folder of plain-text files. What you sell, who your clients are, what is in motion, what is due this week. Plus one instructions file that tells Claude how to be an operator about it: read the memory, brief me, catch what is slipping, ask before assuming. That is the whole trick. No code. Ten minutes. Free Claude account. Comment ATLAS on the pinned post and the system sends it to you. What you get is real: morning rundowns on your actual business, "what am I forgetting" answers that sting a little, a map of your attention. What it cannot do is also real: it has no hands. It cannot log into GoHighLevel, audit a workflow, or fix what it finds. It advises. The version with hands is the tool this community is built around. But start with the brain. It is free, it is yours, and it will show you exactly why hands matter.