@elitedcs/ghl-mcp 3.93.1 → 3.94.1

package/CHANGELOG.md

@@ -1,5 +1,43 @@
 # Changelog
 
+## 3.94.1: Fixes from a full walk-through of the cockpit
+
+**What they bought now comes before the Brief.** Building a client account is 13 steps: after the intake, step 6 asks what your client paid for, then the Brief and the Blueprint Review are written for exactly that. Before, you ticked what they bought at Approve, after the review was written, so the review was out of date on almost every build. If you change what they bought after the review, it says so in one line, and you choose: Approve anyway, or Update the review (this uses Claude). Approving anyway is recorded with the approval. Every step number the cockpit mentions follows the new order.
+
+**Build never builds on top of a different plan's build.** If an earlier build of a different plan left items in the account that this build would reuse, the list before Build now names them and will not let you confirm. It gives you the exact words to ask Claude to undo that earlier build (Claude lists what it would remove and removes nothing until you type DELETE), or you can stop there. Building the same plan again, for example after a failed run, carries on as before. Draft workflows already in the account that no build made are named on the list so you can check them.
+
+**The skills install never writes through a link.** If your Claude skills folder, or a folder in it, is a link to somewhere else, such as your own copy of a skill, GHL Command now leaves it alone and says "not touched, linked elsewhere" when it starts. install_skills says which skills were not installed and why. Before, it could write its own files into the place the link pointed at.
+
+**A step that fails says why.** When Write the Brief or Build stops with nothing to report, its card now says so in one sentence instead of showing the button again with no word. If Claude Code is signed out on your computer, the card says that and how to fix it (type /login in Claude Code, then click the button again). A Brief that stops is said to be safe to run again, because it does not change the account.
+
+**Workflow Audit Pro says what it writes, and tidies its own pipeline card.** The confirmation names the test contact's real address: one made from your email with a new code for each check, still your inbox. If a workflow makes a pipeline card for the test contact during the check, the check now takes that card out of your pipeline when it ends, and only that card, and only when the check made the contact itself. If your form matched the test to a contact already in the account, the check leaves its cards alone and lists them for you to look at. If the check cannot confirm the test contact was its own, it does the same and says so. The confirmation says this before you start, and the report says so plainly if a card could not be taken out. On the report, a text step the test lead had no phone number for is marked Not tested, not Did nothing.
+
+**Smaller fixes.** The daily brief no longer says "this morning" whatever the time, and its money tiles say Not connected or Not recorded when that is what happened, instead of Not read. A list that has never loaded says either that it is loading or why it could not load, never both. The Industry list on a new client shows Other once. On Brands, "Whose site is it?" starts unanswered, and the brand is not pulled until you answer. Opening the handoff, the account detail or the Blueprint Review with no client chosen now shows a proper page with a way back, instead of one bare line. After you save what a client bought, the Blueprint Review panel checks again whether the review is now out of date. Lines taken out of a review for promising more than the build makes are listed in a folded note, each line as it was. A Brands design run no longer shows routine technical log lines or full file paths (lines that say something went wrong are kept), has an Open the result button when it finishes, and its time and cost estimate now includes a faster measured run (a pass is 5 to 13 minutes and $0.85 to $1.75 of your own Claude usage). On the daily brief, an account that is being read again shows as still loading, not as one that could not be read. On Your agency, the reminder to add an agency name goes away as soon as you save one. Pages under For this client that read nothing from the account no longer carry a "Read live from GoHighLevel" heading.
+
+## 3.94.0: Ask your client first, and build only what they chose
+
+**Texting, call answering and reactivation are asked first.** Under What they bought, a new box asks you to confirm three things with your client before Build: can this business text leads, how should calls be answered (an AI answering live, an AI replying by text or chat and booking, or a text back after a missed call), and who reactivation should reach, with which offer and by email or by email and text. Nothing is pre-selected. When your client's own form already answered texting, it shows that answer with a Change link. The answers save with Save what they bought, and changing one after the plan is approved means approving it again. In this version only the text back after a missed call, and reactivation for leads who go cold from now on, are built from these answers. An AI answering the phone live, an AI replying by text or chat, and reactivation for an existing or imported list are recorded but not built yet; they read as coming in a later update.
+
+**A question left blank builds nothing for that item.** It shows as Waiting for your client's answer, on the jobs list and on the client document, and never as built. Before this, a build with texting unanswered made no texts but still made the text templates and asked for a phone number and text registration "for the texts".
+
+**Built means built.** Before Build, each item says Will build. After it, an item says Built only when the build really made it. The client document never promises "No call goes unanswered"; it says what was done for calls.
+
+**The Blueprint Review keeps to what this build makes.** What we're building is now the build's own list, with anything waiting on your client marked as waiting. The review's own words cover the diagnosis only, and every line is checked twice before it is shown: a line that promises more than the build makes gets one more try, then is taken out. A review written before the answers, the texting answer or what they bought changed says so, and the client document waits until it is written again.
+
+**When your client answers later, Build adds only what was waiting.** Approve the updated plan and run Build again: everything already made is kept and nothing is made twice.
+
+**A first-time missed caller gets a pipeline card.** When Missed Call Text-Back is built, a first-time missed caller gets a card at the first stage of the pipeline, with an owner. A caller who already has a card does not get a second one.
+
+**Instructions name the message, not the channel.** The review link and the quote and review checks say "message" or "request", because with texting off the email is what goes out.
+
+**Two client pages read something real.** Prospect demo shows the demo bots and knowledge bases on the account. Sign-off and handoff shows what is built, what waits on your client and the jobs still open, from the build record.
+
+**The guide says Google Calendar connect is not available yet,** matching the Connections card.
+
+**The README is corrected.** HighLevel's own API and MCP have built workflows since October 19, 2026, so the README no longer says GHL Command is alone in that. Module tool counts now match the code.
+
+**Releases test faster.** The release tests run in six parts at once, and nothing publishes unless every part passes.
+
 ## 3.93.1: One build at a time per account
 
 **A build can no longer run twice at once on the same account.** If a second build starts for an account while one is still writing, it stops at once, says a build is already running for that account, and writes nothing. Before this, two builds running together could each create every workflow and email, leaving two of each. The build step is also told to start the build once and wait for its answer.

package/README.md

@@ -10,7 +10,7 @@ Works with **both the Claude Desktop App and Claude Code terminal** — your cho
 
 **Try it FREE (read-only tier).** Grab an instant free key at [ghlcommand.com/free](https://ghlcommand.com/free?utm_source=npm&utm_medium=readme) — 112 read-only tools on your own GHL account once the two-step install is finished (the same install as the paid tier — upgrading later is just a license swap): the account-wide workflow audit (`audit_workflows` finds the silently broken references GHL never warns you about), account health, and every read across contacts, pipelines, funnels, and workflows. Write tools stay visible and answer with upgrade info. Buying later with the same email upgrades your key in place — 1 machine on free, 3 when you upgrade.
 
-**How this differs from HighLevel's official MCP.** HighLevel ships an official Anthropic MCP that reads and writes your CRM (contacts, conversations, calendars, opportunities, payments, social) over the public API — it's free and it's good; use it for that. What it *cannot* do, because the public API doesn't expose it: **build or edit workflows, build funnels/landing pages, or build forms.** That's exactly what GHL Command does — the programmatic Workflow Builder, funnel/page builder, form builder, deep workflow cloning, the silent-failure `audit_workflows`, and Blueprint (build a whole client account from one intake — ships as a guided skill that **installs itself automatically** with this package; nothing to download), all on GHL's internal APIs. **Official MCP = read/write your CRM. GHL Command = build and fix it.** They stack fine together.
+**How this differs from HighLevel's official MCP.** HighLevel ships an official Anthropic MCP that reads and writes your CRM (contacts, conversations, calendars, opportunities, payments, social) over the public API. It's free and it's good; use it for that. Since October 19, 2026 HighLevel's public API, and its MCP, can also create, validate, update, publish and delete workflows. GHL Command builds a different way: it works through GoHighLevel's own Workflow Builder (which is why setup includes logging into GoHighLevel once), and it adds the funnel/page builder, form builder, deep workflow cloning, the silent-failure `audit_workflows`, and Blueprint (build a whole client account from one intake; it ships as a guided skill that installs itself automatically with this package, nothing to download). If GoHighLevel changes its builder, those tools can stop working until we ship a fix. They stack fine together.
 
 **Bundled skills (install automatically).** The package ships guided skills that teach Claude complete multi-step workflows on top of the raw tools. On every start the server installs/updates them into your `~/.claude/skills/` (never overwriting files you've edited — your version wins and is reported). Current skills: **Blueprint** — turn one client intake into a complete reviewed GHL account build; **Clone Site** — point at a live URL and get a rebranded, deploy-ready copy for a client, with a recorded rights declaration, a style-only lane when you have no rights, and a mandatory pre-launch report of every lead/booking/payment destination, borrowed testimonial and inherited claim still belonging to the original owner; **GHL Reports** — counts, lists, and weekly reports that come back fast and verified (one-call windowed contact counts via `get_contact_count`, clean CSV list recipes, composed account summaries — never a guessed number, works on the free tier). Verify or repair any time by asking Claude to `run install_skills`, then fully restart Claude so new skills load.
 
@@ -34,7 +34,7 @@ Built by [Elite DCs, LLC](https://elitedcs.com).
 > "Thank you, Jerry. You're a star."
 > — Frankie B.
 
-30+ releases on npm since launch (v3.6.0 → v3.34.0). Bugs get fixed in days, not quarters.
+170+ releases on npm since launch (v3.0.0 onward). Bugs get fixed in days, not quarters.
 
 ---
 
@@ -44,7 +44,7 @@ Most GHL integrations give you read access and basic CRUD. This server goes far
 
 | Exclusive Feature | Why It Matters |
 |---|---|
-| **Programmatic Workflow Builder** | The only tool on the market that can create, edit, and publish GHL workflows via code. GHL's public API has no workflow write endpoints — this uses their internal backend API (the same one their UI calls) to give Claude full workflow CRUD. Build entire automation sequences from a conversation. |
+| **Programmatic Workflow Builder** | Create, edit, and publish GHL workflows from a conversation, through GoHighLevel's own Workflow Builder (the same backend its screen uses), with every action in the exact shape that builder saves. HighLevel's public API has had workflow endpoints too since October 19, 2026. Build entire automation sequences from a conversation. |
 | **Deep Workflow Cloning** | Clone any workflow with full UUID remapping — every action, trigger, branch, and internal reference gets new IDs. No other tool can duplicate a workflow programmatically without ID collisions. |
 | **Funnel & Page Editor** | Read and edit funnel page content — sections, elements, rows, CSS, tracking codes. Goes beyond the public API's list-only access. |
 | **Form Builder** | Full form CRUD including field management, conditional logic, and auto-responder configuration via internal API. |
@@ -149,7 +149,7 @@ Approve the tool call. Server validates your license, verifies your GHL credenti
 "What calendars do I have?"
 ```
 
-Full step-by-step setup with screenshots: [elitedcs.com/ghl-mcp-setup-guide.html](https://elitedcs.com/ghl-mcp-setup-guide.html)
+Full step-by-step setup: the guide ships inside this package as `guide/guide.html`, and members find the same guide in the Members Hub.
 
 ---
 
@@ -421,7 +421,7 @@ Highlights (full guide: [`docs/HEADLESS.md`](docs/HEADLESS.md)):
 | `update_webhook` | Update a webhook |
 | `delete_webhook` | Delete a webhook |
 
-### Documents (4 tools)
+### Documents (6 tools)
 
 | Tool | What It Does |
 |---|---|
@@ -429,6 +429,8 @@ Highlights (full guide: [`docs/HEADLESS.md`](docs/HEADLESS.md)):
 | `get_document` | Get document details and signature status |
 | `delete_document` | Delete a document |
 | `send_document` | Send a document for e-signature |
+| `list_document_templates` | List your reusable contract templates |
+| `send_document_template` | Create a contract for a contact from a template; a draft unless you ask it to send |
 
 ### Payments (4 tools)
 
@@ -490,7 +492,7 @@ Highlights (full guide: [`docs/HEADLESS.md`](docs/HEADLESS.md)):
 | `export_account` | Full sub-account backup — contacts, pipelines, workflows, funnels, forms, fields, tags, calendars, users |
 | `compare_locations` | Side-by-side diff of two sub-accounts for auditing or migration |
 
-### Location Switcher & Token Registry (6 tools)
+### Location Switcher & Token Registry (9 tools)
 
 | Tool | What It Does |
 |---|---|
@@ -500,10 +502,13 @@ Highlights (full guide: [`docs/HEADLESS.md`](docs/HEADLESS.md)):
 | `register_location` | Add a sub-account and its API key to the token registry |
 | `unregister_location` | Remove a sub-account from the token registry |
 | `list_registered_locations` | List all sub-accounts stored in the token registry |
+| `register_agency_key` | Save your agency-level Private Integration key (needed for agency writes such as adding a user) |
+| `register_company_firebase` | Save another company's builder login so the workflow builder works in its sub-accounts |
+| `unregister_company_firebase` | Remove a company's saved builder login |
 
 ### Workflow Cloner (1 tool)
 
-> Deep-clones a workflow with full UUID remapping. Every action, trigger, branch, and internal cross-reference gets a new unique ID. GHL has no built-in way to duplicate a workflow programmatically — this is the only tool that can.
+> Deep-clones a workflow with full UUID remapping. Every action, trigger, branch, and internal cross-reference gets a new unique ID. GoHighLevel's screen can copy a workflow by hand; this does it from a conversation, inside the same sub-account, with every internal reference remapped.
 
 | Tool | What It Does |
 |---|---|
@@ -533,11 +538,11 @@ Highlights (full guide: [`docs/HEADLESS.md`](docs/HEADLESS.md)):
 |---|---|---|
 | **Social Planner** | 5 | Create, list, delete posts, view connected accounts |
 | **Businesses** | 5 | Full CRUD on business entities |
-| **Blogs** | 5 | Posts, authors, categories, URL slugs |
+| **Blogs** | 6 | Posts, authors, categories, URL slugs, and a check that a slug is free |
 | **Funnels** (public API) | 2 | List funnels and their pages |
 | **Forms** (public API) | 2 | List forms and view submissions |
 | **Surveys** | 2 | List surveys and view submissions |
-| **Users** | 2 | List and view team members |
+| **Users** | 3 | List, view and add team members (adding needs the agency key) |
 | **Media** | 2 | Browse and delete media files |
 | **Campaigns** | 1 | List campaigns |
 | **Workflows** (public API) | 1 | List all workflows |
@@ -899,7 +904,7 @@ Built by **[Elite DCs, LLC](https://elitedcs.com)** — a digital marketing and
 
 **Tech stack:** TypeScript, Node.js, esbuild, MCP SDK, Zod, GHL API v2, Firebase Auth
 
-**Version:** 3.93.1
+**Version:** 3.94.1
 
 ---