AI builders wrestle with fragmented MCP tool connections
AI agent and automation builders have trouble connecting agents to MCP tools consistently: they run into missing OAuth support, multi-server setup friction, and a lack of standardized client support. Some resort to custom proxies or scripting, while choosing the right server set for a task can remain manual. The signals point to a shared integration problem, but not to one universally missing feature.
For AI agent builders and self-hosting automation users. Mentioned from Nov 2024 to Mar 2026 on product forums, GitHub and Hacker News.
4 different people described this problem in 4 separate discussions.
- Indie fit
- 7.0/10
- Pain
- 6.5/10
- Frequency
- 5.8/10
- Willingness to pay
- 5.0/10
- Momentum
- 5.0/10
- Who pays
- Professionals
- Competition
- Medium
- Build difficulty
- Medium
What people said
Quoted word for word. Follow a link to read the whole discussion.
The workaround I have been using is running a lightweight OAuth proxy in front of the n8n webhook — basically a small Express server that handles the OAuth flow and forwards authenticated requests to the n8n webhook URL. It works but it is annoying overhead that should not be necessary
The ideal solution would be an official Home Assistant-supported method that simplifies this process and allows users to easily connect their conversation agent to multiple MCP servers.
Build brief
See what to build and who will buy it
- 2 product ideas with the smallest useful version and pricing
- 5 places to find your first customers
- 2 more quotes from people who have this problem
- Current workarounds, existing solutions and risks