Weavz MCP Server

Governed app access for AI agents — one hosted runtime exposing 1,000+ apps (12,000+ tools) through Code Mode MCP, with Human Gates approvals, scoped per-user credentials, sandbox, files, and durable state.

View on weavz.io

Documentation

Agent integration infrastructure for SaaS and AI teams

Build agents that can safely act across customer apps.

Use the API or SDK to provision workspaces, add users, generate connect links, create MCP servers, issue scoped tokens, and plug governed app access into your own agent harness.

Build with WeavzTry in an AI client

Book a demo/Read docs/Start free

1,000+ apps12,000+ agent toolsScoped per workspace and end user

Builder provisioning loop

Create a customer workspace, connect their apps, then hand your agent a scoped MCP endpoint.

POST /workspaces

POST /end-users

POST /connect-token

POST /mcp/servers

Gmail

Calendar

Slack

HubSpot

Notion

Salesforce

GitHub

Sheets

Provision workspace

Create one scoped environment per customer, team, project, or agent.

Add aliases

Mount Slack, Gmail, files, state, browser, and custom APIs as stable names.

Connect each user

Generate hosted connect links for OAuth, API keys, and customer-owned accounts.

Hand off to agents

Issue MCP endpoints and per-user tokens to your harness, then audit every action.

Start from the AI client you already use

Try in ClaudeTry in ChatGPTOther agents

Try one governed Weavz workspace from Codex, Claude Code, Cursor, VS Code, Cline, Gemini CLI, OpenCode, custom remote MCP clients.

Platform coverage

0+Appsbusiness systems agents can reach

0+Agent toolsactions and triggers on demand

0+Triggerswebhooks, app events, and polling

0Built-insfiles, state, browser, code, memory

The apps people already expect agents to use.

Connect Gmail, Slack, calendars, CRMs, docs, tickets, finance tools, and internal APIs. Weavz turns each configured workspace into governed app access for chats, copilots, workflows, and your own product.

Slack

Google Sheets

Gmail

Google Calendar

Discord

GitHub

Notion

OpenAI

Linear

HubSpot

Airtable

Asana

Figma

Twilio

Telegram

Shopify

Salesforce

Dropbox

Google Contacts

Jira

Trello

ClickUp

Todoist

Monday.com

Google Drive

Google Forms

Google Tasks

Microsoft Teams

Intercom

Typeform

HTTP / REST

Stripe

Mailchimp

Zendesk

SendGrid

Outlook Calendar

OneDrive

Google Analytics

Instagram

Facebook Pages

YouTube

Spotify

Zoom

Freshdesk

Pipedrive

WordPress

Webflow

Supabase

Claude

Google Gemini

1,000+ apps, 10,000+ actions, and 12,000+ agent tools. Don't see yours? Tell us what you need next →

Agent tool workspaces

Start with a ready workspace, then tune the controls.

Weavz creates curated tool workspaces for common agent jobs. Each one adds the right apps, keeps credentials scoped to each user by default, and can expose the workspace through governed MCP access.

Open tool workspacesSet up your own

Everyday agent workspace

Browser, computer, files, memory, email, calendar, chat, and projects

View workspace

Founder command center

Email, calendar, CRM, docs, chat, and project follow-through

View workspace

Engineering agent

GitHub, issues, docs, chat, files, and scoped state

View workspace

Support escalation desk

Tickets, customer replies, billing context, and approvals

View workspace

Marketing growth studio

Campaigns, CRM context, launch lists, email, and team updates

View workspace

Data ops workbench

Databases, tables, spreadsheets, HTTP enrichment, and files

View workspace

Human Gates demo desk

Approval-required flows for sends, shared writes, and MCP calls

View workspace

Your stack

Create a custom workspace

Choose your own connectors, aliases, credentials, MCP access, and approval rules for the agent job you want to run.

Set up your own

Provision customer agent workspaces

Create workspaces, users, connect links, MCP servers, and scoped tokens from your backend.

Start here

Plug into your agent harness

Use Weavz with OpenAI Agents SDK, LangGraph, CrewAI, ADK, Vercel AI SDK, and custom MCP clients.

Start here

Try it directly in an AI client

Add Weavz Agent Workspace to Claude, ChatGPT, Codex, Cursor, and other remote MCP clients.

Start here

What Weavz controls around every action.

Agents do not need a bigger pile of tools. They need an environment that knows who is acting, what they can touch, what needs review, and what state should survive the run.

Connect users and apps

Hosted Connect lets each person authorize the accounts their agent should use. Workspaces keep app aliases, connection strategy, and tenant boundaries clear from the first run.

Connect Slack

Click to authorize

AC

Alice

BK

Bob

CW

Carol

Expose approved actions to agents

Use Tool Mode for focused surfaces or Code Mode for broad workspaces. Agents search, read typed APIs, and execute without loading thousands of schemas up front.

Agent

read

execute

Weavz MCP Server

Context100%

Govern risk, state, and audit

Human Gates pause sensitive work. Filesystem, State KV, memory, browser, and Sandbox tools let agents finish multi-step jobs with a record of what happened.

github.create_issue

Approval required

Redacted request

Reviewer sees enough context

Approved

Workflow resumes

DashboardHosted linkWebhook

Agent Browser

Logged-in UI work

Sandbox

Python, Shell, and JS

Filesystem

Files and outputs

State KV

Cursors and checkpoints

one workspace policy, scoped per user or tenant

Four ways in. Same app actions on demand.

Builders provision through API and SDKs. Agents use MCP. Operators use the CLI or dashboard when they need direct control.

API / SDKs

Provision workspaces, integrations, end users, connect links, MCP servers, and scoped tokens from your backend. TypeScript and Python SDKs keep action inputs typed and validated.

MCP Servers

Expose mounted workspace integrations as an agent execution layer. Tool Mode for focused setups. Code Mode for production: 3 meta-tools, typed APIs, parallel calls, Filesystem, State KV, and Sandbox execution.

CLI

Install Weavz in a terminal, SSH session, container, or coding agent. Device login gives the CLI a scoped MCP token for the active workspace.

Dashboard

Point-and-click setup. Connect integrations, configure MCP servers, test actions in the Playground, review Human Gates, manage end users, and tune workspace settings.

API / SDKsMCP ServersCLIDashboard

const { workspace } = await client.workspaces.create({
  name: "Acme Agent Workspace",
  slug: "acme-agent-workspace"
});

const { endUser } = await client.endUsers.create({
  workspaceId: workspace.id,
  externalId: "user_456"
});

const { server, mcpEndpoint } = await client.mcpServers.create({
  workspaceId: workspace.id,
  mode: "CODE",
  authMode: "oauth_and_bearer"
});

Simple, transparent pricing

Start free, scale as you grow. No hidden fees.

Free

$0/month

For exploring agents, actions, and state

  • 20,000 actions/month
  • 10 connections
  • 25 MCP servers
  • Community support

Get Started Free

Pro

$29/month

For launching production MCP workflows

  • 200,000 actions/month
  • 100 connections
  • 500 MCP servers
  • Email support

Start Pro Trial

Team

Most Popular

$129/month

For teams shipping embedded workflow products

  • 1,000,000 actions/month
  • 500 connections
  • 5,000 MCP servers
  • Priority support

Start Team Trial

View all plans & compare features →

/mcp/weavz

Give your agent a workspace that can act.

Start free, add Weavz to an AI client, or walk through the production setup with our team.

Start freeAdd to AI clientBook demo