Claude Agent Team

More

Name	Category	Notes	Added	Actions
No keys stored yet

Key Name *

Key Value *

🔑

Files

Loading…

🧠 OpenRouter live from API

Loading…

Per account key (lifetime used / cap)

⚡ Glitch Connect month-to-date

Loading…

Per sub-account (spent / monthly cap)

📊 Claude Max (5x) Usage Tracker

Real-time monitoring of your Claude Max plan usage

Plan

Max 5x

🔑 Total Usage This Month 0%

0%25%50%70%85%95%100%

Total Requests

0

Today

0

This Week

0

Est. Cost

$0

📊 Usage by Agent

Loading...

👤 Usage by Task Type

Loading...

🚨 Alerts & Warnings

✅ All usage levels normal. No alerts.

Alert Thresholds

25%

Info

50%

Heads Up

70%

Warning

85%

High

95%

Critical

⏱ Recent Activity (last 20)

Loading...

📈 Daily Usage Trend (Last 14 Days)

Loading...

Title *

HighLevel MCP Server

● Live

AI bridge to your GHL data — query, automate, and orchestrate everything in your HighLevel account via Model Context Protocol. 36 native MCP tools + 834 BusyBee HTTP tools (691 working). Per the GHL-independence strategy: Glitch owns the brain — GHL is the swappable system-of-record reached through these stable APIs.

✅ Reviewed 2026-06-11 · canonical reference → Systems Manual v6.1 (Notion)

Full Documentation → Python Client Helper Library GHL Developer Docs

36

Tools Available

✓

Connection

8

MCP Endpoint Configuration

{
  "mcpServers": {
    "ghl-mcp": {
      "url": "https://services.leadconnectorhq.com/mcp/",
      "headers": {
        "Authorization": "Bearer ••••••••-••••-••••-••••-••••••••••••",
        "locationId": "ifitVy09cFwlEVgEAFDS"
      }
    }
  }
}

n8n Integration Config

{
  "MCP URL": "https://services.leadconnectorhq.com/mcp/",
  "Transport": "HTTP Streamable (n8n v1.104+)",
  "Auth": "Header Auth → Authorization: Bearer [REDACTED - use vault]",
  "System Prompt": "The location id is ifitVy09cFwlEVgEAFDS"
}

GHL Automation Webhook Body

Standard webhook payload for native GHL workflow automations (→ n8n/Make). Note: Agent Studio is retired — reactivity now comes from the Glitch engine + the polling watcher (Systems Manual v6.1 §3):

{
  "contactId": "{{custom_values.contact_id}}",
  "locationId": "{{custom_values.location_id}}",
  "pit_token": "{{custom_values.ghl_pit}}",
  "tags": "{{custom_values.contact_tags}}",
  "website": "{{contact.website}}",
  "social_media_links": "{{custom_values.social_media_links}}"
}

All 36 Native MCP Tools (+ 834 BusyBee HTTP tools per tenant — 691 working)

#	Tool	Endpoint	Description

Required PIT Scopes

What You Can Do

Contact Management on Autopilot

Fetch, update, create, tag & segment contacts via AI prompts. Upsert contacts — AI figures out create vs update.

Conversational AI Messaging

Send messages to any conversation, search by keywords/status, auto-reply to common questions via MCP + chatbots.

Custom Multi-Step Workflows

Chain MCP actions: on new contact → tag, send welcome message, add to pipeline — all via AI agents.

Payment & Transaction Intelligence

Fetch orders & transaction history on demand. Build AI dashboards for payment analysis and refund flagging.

Social Media Posting

Create, edit, and retrieve social media posts. Get analytics across connected accounts.

Blog Management

Create and update blog posts, check slug availability, manage categories and authors — all via AI.

⚡

Glitch Mega-Skill

GHL MCP OpenRouter Apify 🏪 Marketplace Listed

Unified webhook skill combining BusyBee MCP (834 GHL tools — 691 working), OpenRouter (LLM + image + video gen), Apify (web scraping), multi-turn memory, and auto-provisioning into one API.

✅ Reviewed 2026-06-11 · canonical reference → Systems Manual v6.1 (Notion)

Full Documentation → Python Source Bot Prompt Webhook Config

36

GHL Tools

834

BusyBee GHL Tools

Sonnet 4.6

Brain (OpenRouter / BYO)

16

Endpoints

1

Unified API

API Endpoints

Base: https://agents.realtalkha.com/api/webhooks/glitch

Method	Path	Action	Description

⚡ Glitch Fleet Operations

● Loading fleet… One Glitch engine · two front doors 84+ connectors · GHL-independent senses Portable onboarding (no snapshot)

Live tenant registry, payment links, webhooks, and the provisioning flow. Heartbeats + polling watcher per account; live health lives in the Operator → Fleet Monitor.

📖 Systems Manual v6.1

🛰️ Live fleet health → Operator · Fleet Monitor

Per-tenant health verdict · triggers · heartbeats · watcher · Glitch Connect spend vs cap · brain cost (30d)

→

Active Tenants

6 tenants

Loading tenants from /api/tenants…

How tenants get here (canonical path — Jun 12 2026): Stripe/welcome → owner installs the marketplace app via /oauth/chooselocation → OAuth callback stores access+refresh tokens → portable orchestrator provisions: scoped PIT · per-tenant BusyBee engine on port 91xx behind nginx /ghl/{locationId}/ (root PM2) · per-account OpenRouter key + cap · Glitch GHL user + AgentMail + identity custom values → owner creates their account in the Glitch app (primary front door; webchat / Telegram / WhatsApp / Discord all optional channels) → status live. Auth model: PIT for ~95% of ops; Glitch user sign-in for identity stamping + UI-only browser actions; AgentMail auto-catches the 2FA code.

🚧 Blockers Checklist 0

Tasks requiring manual human action — tagged [HUMAN TASK] by agents.

Hide completed

Loading blockers...

👤 Human Tasks

Manual tasks assigned by AI agents. Chat, upload files, and send completed work back to agents.

0

Total

0

Pending

0

In Progress

0

Completed

0

Urgent

Loading human tasks...

Title *

Description

Assignee

Priority

Task Chat

📎 Attachments + Upload File

No attachments yet

No messages yet. Start the conversation.

📤 Send Work Back to Agent

Target Agent *

Summary Message

🔔 Account Notifications

Everything that needs attention — bot escalations, AI caps, billing, auth.

High: 0 Showing: 0

Loading notifications…

📋 Integration Cheat Sheet

Copy-paste ready credentials, endpoints, and configs for all integrations. Click any block to copy.

✅ Reviewed 2026-06-11 · canonical references → Systems Manual v6.1 (Notion) · New Offer Ladder (Notion)

⚡ OpenRouter / Apify / BusyBee / Glitch Connect

Direct providers + Glitch Connect (white-labeled multi-vendor gateway — 700+ APIs / 1,000+ skills, included in every plan, $100/mo per-account credit cap; BYO keys exempt). See Systems Manual v6.1 §3.

Copied!

OpenRouter Base URL

https://openrouter.ai/api/v1

Copied!

Glitch Connect Base URL (key in .env / per-tenant vault)

https://api.heybossai.com/v1

Copied!

Glitch Connect Image Gen (proven path — images return base64 in message.images)

POST https://api.heybossai.com/v1/chat/completions {"model":"google/gemini-3-pro-image-preview","messages":[{"role":"user","content":"YOUR PROMPT"}]}

Copied!

BusyBee MCP Endpoint

https://agents.realtalkha.com/ghl/{location_id}/tools/call

Copied!

OpenRouter Chat Completions

POST https://openrouter.ai/api/v1/chat/completions

Copied!

Apify Actor Run Endpoint

POST https://api.apify.com/v2/acts/{actor}/runs

Copied!

Auth Headers

OpenRouter: Authorization: Bearer YOUR_OPEN_ROUTER_API_KEY Apify: Authorization: Bearer YOUR_APIFY_API_KEY BusyBee: POST body with {"name":"tool_name", "arguments":{...}}

Copied!

Chat Example (cURL) — OpenRouter

curl -X POST https://openrouter.ai/api/v1/chat/completions \ -H "Authorization: Bearer YOUR_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"anthropic/claude-sonnet-4-6","messages":[{"role":"user","content":"Hello"}]}'

Copied!

Image Gen Example (cURL) — OpenRouter Flux

curl -X POST https://openrouter.ai/api/v1/chat/completions \ -H "Authorization: Bearer YOUR_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"black-forest-labs/flux-schnell","messages":[{"role":"user","content":"a sunset over mountains"}]}'

Copied!

Popular Models Quick Reference

🧠 Glitch Reasoning Model (OpenRouter)

Copied!

Base URL

https://openrouter.ai/api/v1

Copied!

Model Name

anthropic/claude-sonnet-4-6

Copied!

Chat Completions Endpoint

POST https://openrouter.ai/api/v1/chat/completions

Copied!

cURL Example

curl -X POST https://openrouter.ai/api/v1/chat/completions \ -H "Authorization: Bearer YOUR_OPENROUTER_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"anthropic/claude-sonnet-4-6","messages":[{"role":"user","content":"Explain quantum computing"}],"temperature":0.7}'

Copied!

OpenAI SDK (Python)

from openai import OpenAI client = OpenAI(base_url="https://openrouter.ai/api/v1", api_key="YOUR_OPENROUTER_KEY") resp = client.chat.completions.create( model="anthropic/claude-sonnet-4-6", messages=[{"role": "user", "content": "Your prompt here"}] ) print(resp.choices[0].message.content)

Notes

Claude Sonnet 4.6 via OpenRouter — the default Glitch brain (model centrally set in glitch_engine_config). Per-tenant capped OpenRouter keys; owners can BYO LLM (OpenAI / Anthropic / Gemini / Perplexity) via Bot Safety → Settings. OpenAI-compatible — drop-in replacement, just change base_url.

🔄 n8n Workflows

Copied!

n8n Instance URL

https://realtalkha.app.n8n.cloud

Copied!

Webhook Base URL

https://realtalkha.app.n8n.cloud/webhook/

Copied!

Test Webhook Base URL

https://realtalkha.app.n8n.cloud/webhook-test/

Copied!

n8n API Base

https://realtalkha.app.n8n.cloud/api/v1

Copied!

Active Webhook Paths

/webhook/telegram-glitch — Telegram inbound listener /webhook/glitch-provision — Marketplace install handler /webhook/glitch-web-scrape — Web scrape handler /webhook/glitch-a2p-compliance-v2 — A2P compliance pipeline /webhook/glitch-stripe — Stripe event handler

Copied!

n8n MCP Tools (Claude Desktop)

search_workflows — Find workflows by name/tag get_workflow_details — Get full workflow JSON create_workflow_from_code — Create from SDK code update_workflow — Update existing workflow execute_workflow — Trigger a workflow run get_execution — Check execution status validate_workflow — Validate before creating search_nodes — Find node types get_node_types — Get node parameter schemas get_suggested_nodes — Curated recommendations archive_workflow — Archive a workflow publish_workflow — Activate a workflow unpublish_workflow — Deactivate a workflow

🏢 GoHighLevel (GHL)

Copied!

API v1 Base URL

https://rest.gohighlevel.com/v1/

Copied!

API v2 Base URL

https://services.leadconnectorhq.com

Copied!

Agency PIT Token

[REDACTED - use vault]

Copied!

Snapshot PIT Token (legacy agency path)

[REDACTED - use vault]

Copied!

Snapshot Location ID (legacy — portable onboarding needs NO snapshot)

xyeL7GNvNh0QEzCnOBKJ

Copied!

BizDev Location ID

ifitVy09cFwlEVgEAFDS

Copied!

BizDev PIT Token

[REDACTED - use vault]

Copied!

Auth Header Format

Authorization: Bearer pit-XXXXX

Copied!

GHL MCP Tools (native MCP server)

calendars_get-calendar-events contacts_get-contact contacts_create-contact contacts_update-contact contacts_upsert-contact contacts_add-tags contacts_remove-tags contacts_get-contacts conversations_search-conversation conversations_get-messages conversations_send-a-new-message locations_get-location locations_get-custom-fields opportunities_search-opportunity opportunities_get-pipelines opportunities_update-opportunity payments_get-order-by-id payments_list-transactions

Copied!

cURL Example — Get Contact

curl -X GET "https://services.leadconnectorhq.com/contacts/CONTACT_ID" \ -H "Authorization: Bearer pit-XXXXX" \ -H "Version: 2021-07-28"

🔗 Webhooks & Endpoints

Copied!

Droplet 1 — ClaudeBotz (factory + existing tenants)

https://agents.realtalkha.com · 142.93.60.203 · FastAPI :8000 · /opt/claudebotz

Copied!

Droplet 2 — Third Floor (⭐ ALL NEW tenants land here)

https://3rd.realtalkha.com · 157.230.232.29 · FastAPI :8090 · /opt/thirdfloor Postgres 16 + Redis 7 · BRAIN_MODE=glitch_engine → calls Droplet 1's engine Status: built + brain proven · go-live wiring pending (channels, Apps A+B, APP_ENV=production)

Copied!

Agent API Endpoints

POST /api/onboard/glitch-setup — ⭐ CANONICAL onboarding orchestrator ({locationId, pit} → full tenant; secret-gated) POST /api/specialist/invoke — Glitch brain (auth ENFORCED: internal token / chat token / settings token) GET /api/project-tasks — Global task feed (status/assignee filters) GET /api/usage/providers — OpenRouter + Glitch Connect usage dashboards POST /api/tasks — Create a task GET /api/tasks/{id}/stream — Stream live output (SSE) GET /api/tasks/{id}/live — Poll task status POST /api/tasks/{id}/action — Send action to agent POST /api/tasks/{id}/comments — Add comment POST /api/tasks/{id}/files — Attach file POST /api/tasks/{id}/telegram — Send Telegram notification GET /api/projects — List projects GET /api/memory/{ns}/{key} — Read agent memory GET /api/kb/search?q= — Search knowledge base POST /api/files — Upload file GET /health — Health check POST /api/busybee/provision — Provision BusyBee per-tenant POST /api/ghl-contact/update — Update GHL contact POST /api/replicate/image-async — Async image generation POST /api/replicate/video-async — Async video generation GET /api/telegram-history — Get Telegram history POST /api/telegram-history/append — Append to history POST /api/telegram-history/clear — Clear history

Copied!

Telegram Bot

Bot tokens stored securely in Vault — use /api/vault to access. Chat ID: [REDACTED - use vault] API: Use vault-stored token with https://api.telegram.org/bot[TOKEN]/sendMessage

Copied!

Workspace File URLs

Files at /opt/claudebotz/workspace/NAME → https://agents.realtalkha.com/workspace/NAME Files at /opt/claudebotz/uploads/NAME → https://agents.realtalkha.com/uploads/NAME

Copied!

n8n Webhook URLs (Full)

https://realtalkha.app.n8n.cloud/webhook/telegram-glitch https://realtalkha.app.n8n.cloud/webhook/glitch-provision https://realtalkha.app.n8n.cloud/webhook/glitch-web-scrape https://realtalkha.app.n8n.cloud/webhook/glitch-a2p-compliance-v2 https://realtalkha.app.n8n.cloud/webhook/glitch-stripe

🛠 Quick Config Blocks

Copied!

n8n HTTP Request Node — OpenRouter Chat

{ "method": "POST", "url": "https://openrouter.ai/api/v1/chat/completions", "authentication": "genericCredentialType", "headers": { "Authorization": "Bearer YOUR_OPEN_ROUTER_API_KEY", "Content-Type": "application/json" }, "body": { "model": "anthropic/claude-sonnet-4-6", "messages": [{"role": "user", "content": "={{$json.prompt}}"}] } }

Copied!

n8n HTTP Request Node — GHL Get Contact

{ "method": "GET", "url": "=https://services.leadconnectorhq.com/contacts/{{$json.contactId}}", "headers": { "Authorization": "Bearer pit-XXXXX", "Version": "2021-07-28" } }

Copied!

n8n HTTP Request Node — OpenRouter (Glitch model)

{ "method": "POST", "url": "https://openrouter.ai/api/v1/chat/completions", "headers": { "Authorization": "Bearer YOUR_OPENROUTER_KEY", "Content-Type": "application/json" }, "body": { "model": "anthropic/claude-sonnet-4-6", "messages": [{"role": "user", "content": "={{$json.prompt}}"}], "temperature": 0.7 } }

Copied!

GHL Workflow Custom Webhook (native automation → n8n)

{ "url": "https://realtalkha.app.n8n.cloud/webhook/YOUR-PATH", "method": "POST", "headers": { "Content-Type": "application/json" }, "body": { "contactId": "{{custom_values.contact_id}}", "locationId": "{{custom_values.location_id}}", "pit_token": "{{custom_values.ghl_pit}}", "tags": "{{custom_values.contact_tags}}", "website": "{{contact.website}}", "social_media_links": "{{custom_values.social_media_links}}" } }

🛡️ Operator — Approvals & Faculties

The human-approval lane plus the fleet's self-scheduling, policy hooks, and connector registry.

Pending approvals 0

show decided

Loading…

🫀 Crons / heartbeats — scheduled autonomous runs

Loading…

Schedule a new run:

🪝 Policy hooks — per-account deny / approval

Loading…

Add a hook:

⚡ Triggers — event-driven autonomous runs

Loading…

Add a trigger:

🧩 Integrations — add a tenant's keys

Pick a sub-account, then paste their API key for any connector. It's stored encrypted for that account and the bot's tools light up automatically. (OAuth ones connect via the owner's button.)

Pick a tenant…

🛰️ Fleet Monitor

Per-tenant liveness + senses + spend: health verdict (6h checks) · triggers/heartbeats · Glitch Connect credits vs cap · brain cost (30d).

Loading…

💰 AI Cost per Tenant

Model-inference spend per sub-account (for rebilling margin). Metadata-only trace logs — no prompts/PII.

Loading…

📈 Skill Attach-Rate

Is the auto-inject working? Skills surfaced (injected) vs actually used. “Served but unused” = retrieval matched it but the bot ignored it (improve its description).

Loading…

🕑 Time Zone — a tenant's date/time

Glitch's current time for this tenant: …

🧩 MCP Servers — remote tools for a tenant

Connect an https MCP server for this tenant. Tool calls are approval-gated; SSRF-guarded.

Pick a tenant…

📄 Forms — a tenant's fillable PDFs

Pick a tenant…

🧠 Memory — a tenant's saved facts

Facts the bot remembers for this account. Add, edit, or remove them.

Pick a tenant…

📚 Knowledge Base — docs the bot reads

Upload a file (PDF/DOCX/TXT/CSV) or paste text. The bot searches these before asking.

Pick a tenant…

🖼️ Media — files the bot can use

Logos, headshots, listing photos the bot can attach when posting/sending.

Pick a tenant…

📘 Skill Library

vetted playbooks every bot can use · owners toggle on/off per account

Loading…

🧠 Proposed skills

Glitch's self-learned drafts from successful sessions — NOT active until you approve them here.

Loading…

Add a playbook (applies to every bot, all dashboards):

🛡️ Guardrails — how careful each bot is

How careful should this bot be?

Loading…

🕘 Recent changes — what changed, with one-click undo

Loading…

📋 Tasks

Every task across all projects, what-needs-you first. Tap a card to open that task's chat.

0 Total Tasks

Done 0

Working 0

Blocked 0

Paused 0

Handoff 0

Overall

0%

|

⚡ Needs attention

Download Open

Overview

Agents

Files

🔴 Live dispatches (this session)

📅 Scheduled Tasks

Projects

📚 Knowledge Base

💰 Expenses & Income

🧾 Recurring Bills & Subscriptions

💵 Products & Sales

Product catalog

Recent sales (Stripe)

Client subscriptions by sub-account (GHL)

API Key Vault

🔐 Secure Vault

Files

🧠 OpenRouter live from API

⚡ Glitch Connect month-to-date

📊 Claude Max (5x) Usage Tracker

📊 Usage by Agent

👤 Usage by Task Type

🚨 Alerts & Warnings

⏱ Recent Activity (last 20)

📈 Daily Usage Trend (Last 14 Days)

HighLevel MCP Server

MCP Endpoint Configuration

n8n Integration Config

GHL Automation Webhook Body

All 36 Native MCP Tools (+ 834 BusyBee HTTP tools per tenant — 691 working)

Required PIT Scopes

What You Can Do

Glitch Mega-Skill

API Endpoints

Usage Examples

GHL MCP Tools (36 native MCP + 834 BusyBee HTTP)

Provider Capabilities (OpenRouter + Apify + BusyBee + Glitch Connect)

n8n & GHL Workflow Integration

n8n HTTP Request Node

GHL Custom Webhook Action

Example: Lead Enrichment Pipeline

Example: Auto-Content Pipeline

GlitchCommandBot Enhanced Prompt

⚡ Glitch Fleet Operations

🛰️ Live fleet health → Operator · Fleet Monitor

Active Tenants

Webhook & API Endpoints

The onboarding pipeline AUTOMATED Jun 12 2026 · one manual step

Onboarding in progress live · per-step status

Stack

🚧 Blockers Checklist 0

👤 Human Tasks

🔔 Account Notifications

📋 Integration Cheat Sheet

⚡ OpenRouter / Apify / BusyBee / Glitch Connect

🧠 Glitch Reasoning Model (OpenRouter)

🔄 n8n Workflows

🏢 GoHighLevel (GHL)

🔗 Webhooks & Endpoints

🛠 Quick Config Blocks

🛡️ Operator — Approvals & Faculties

Pending approvals 0

🫀 Crons / heartbeats — scheduled autonomous runs

🪝 Policy hooks — per-account deny / approval

⚡ Triggers — event-driven autonomous runs

🧩 Integrations — add a tenant's keys

🛰️ Fleet Monitor

💰 AI Cost per Tenant

📈 Skill Attach-Rate

🕑 Time Zone — a tenant's date/time

🧩 MCP Servers — remote tools for a tenant

📄 Forms — a tenant's fillable PDFs

🧠 Memory — a tenant's saved facts

📚 Knowledge Base — docs the bot reads

🖼️ Media — files the bot can use

📘 Skill Library

🧠 Proposed skills

🛡️ Guardrails — how careful each bot is

🕘 Recent changes — what changed, with one-click undo

📋 Tasks