Map a process, journey, or customer hypothesis on a shared whiteboard.
- Cost
- Free and paid plans; limits vary
- Setup
- Account; browser
Opening your saved workspace. Public pages remain available below.
Bring the right systems into one workflow with clear permissions, ownership, and testable data flows.
Your roadmap0 of 15 checks marked · opening your workspace
Define what should move between systems before authorizing access.
Choose what fits. Alternative platforms are choices, not extra requirements. Use your own authorized accounts.
Map a process, journey, or customer hypothesis on a shared whiteboard.
Maintain a simple experiment log, data dictionary, cost model, or operations report.
A field map and a clear integration boundary.
Fill in the brackets for your selected approach: Connect two apps.
Help me connect [source app] to [destination] for [purpose]. Identify available native integrations first, then a supported connector. Document data direction, fields, scopes, source of truth, and revocation. Start read-only with sample data, then test access boundaries and failures. Do not expose credentials in prompts or source code. Start with the smallest useful version and keep additional services optional. Recommend only relevant tools and verify supported skills or plugins before using them. Explain costs and permissions. Record checks actually performed and unresolved issues. Ask for approval before publishing, sending messages, connecting live accounts, or spending money.
Checks record your progress; they do not execute tools or verify business results. Official resources reviewed November 2026. Check current pricing, license terms, and host compatibility. These guides have not been tested as end-to-end business or integration workflows.
Browse related repositories →Complete this for one proposed connection. Confirm access before promising a feature or building around it.
Loading saved notes… Do not enter credentials or confidential client data. Status: opening your workspace.
0 blocked · 8 items need assessment or follow-up. These are your own labels, not an automated verdict.
A completed worksheet is a feasibility record, not a working integration or a security certification. Exports are created locally; this worksheet does not send notes to an AI or external service.
Expand a topic for courses, documentation, and tools from the supplied learning curriculum. Links already recommended in the guide above are omitted.
An API call is just an HTTP request. If you can send one by hand and read the reply, connecting tools stops being magic.
Requests, responses and status codes explained by the browser makers.
Open-source, local-first API client that stores collections as plain files you can commit to Git.
See exactly what another system sends your way.
Some tools give you a private token for your own account; apps that act for many users use OAuth. Pick the narrowest one that works.
The industry-standard authorization protocol and its flows for web, mobile and device apps.
A practical guide with step-by-step flows and real examples.
Real-world example of a scoped private token you can create, rotate and delete, instead of a full-access key.
REST docs for contacts, messaging, calendars, payments and webhooks in a hosted CRM.
MCP is an open standard that lets an AI app call your tools and read your data. Use trusted existing servers before writing your own.
The official introduction: servers expose tools, resources and prompts that AI apps can use.
How to add, scope and manage MCP servers in Claude Code.
Reference server implementations for learning, with pointers to community servers.
Searchable directory of MCP servers.
When no server exists for your tool, a small MCP server (or an n8n workflow exposed as one) is the cleanest way to give an agent exactly the actions you choose.
A tutorial that builds a small weather server in Python or other languages and connects it to a client.
Where to host it, how users authenticate, size and timeout limits, and how to share it.
Lets n8n act as an MCP server so clients can use its tools and workflows.
Every connector is a door. Learn the known attacks and scope each one to the smallest set of actions.
Confused deputy, token passthrough, SSRF, local server compromise and scope minimization, each with fixes.
Says to write your own servers or use providers you trust, and notes Anthropic does not security-audit MCP servers.
Pick two or three to try. Check the licence, since all three below are reported to be AGPL; hosting it for others can bring obligations.
Modern open-source CRM described as an open alternative to Salesforce, designed for AI; confirm its licence terms in the repo.
Long-established open-source web CRM for people, companies and opportunities.
Open-source CRM that keeps data under your control; large user base.
Fully featured open-source CRM from the Frappe team, licensed AGPL-3.0.
Choose by a short test: can you import contacts, call its API, receive a webhook, and back it up? Read each product's own docs for those four.
Product intro and developer guides.
Setup, administration and developer docs, including extensions.
Install, admin and developer guides for SuiteCRM 7 and 8.
Beginner guide: leads, team, email, follow-ups and converting leads to deals.
Additional curriculum comes from the supplied AI Frontiers playbooks. Resource descriptions and cost labels are source guidance, not a fresh verification of pricing, compatibility, or results. Open the official instructions before using a tool. Checklist marks are your own progress notes.