Replies: 1 comment · 1 reply
|
Thanks for converting this into a discussion. The idea mainly came from experimenting with MCP and thinking about how AI assistants could help developers debug or understand what’s happening inside their Appwrite projects. A lot of useful signals already exist , auth logs, function executions, database schemas but they’re usually spread across different places. I was wondering if exposing a small, read-only MCP layer on top of Appwrite APIs could allow AI tools to reason about project state in a safer and more structured way. For example, an assistant could look at recent auth failures, function errors, or schema changes and help explain what might be going wrong. I’m currently experimenting with MCP in Node.js, so if this direction seems interesting for the ecosystem, I’d be happy to try building a small prototype MCP server that works with Appwrite’s APIs. Curious to hear what the team and the community think about this idea. |
|
Really like this direction. A read-only MCP layer feels like a practical way to make Appwrite project state easier to reason about without increasing write risk. If you prototype it, I think the most valuable first scope would be: recent auth failures (with reason + timestamp) If you share a prototype, I’d be happy to test it and give feedback from a developer workflow perspective. |
Uh oh!
There was an error while loading. Please reload this page.
🔖 Enhancement description
Hi Appwrite team
I’m experimenting with MCP (Model Context Protocol) for exposing backend
capabilities to AI assistants in a controlled way.
Would Appwrite be open to a community MCP server that provides read-only access to:
This could help AI answer questions like:
“Why are auth requests failing?” or “What changed recently?”
I’d love to build this in Node.js if the idea fits.
🎤 Pitch
In my specific case, I’d like to see AI assistants safely explore and reason about Appwrite project state (auth activity, database schemas, function executions) without needing to directly access internal services or bypass security constraints.
A Model Context Protocol (MCP) server could offer a structured read-only API that exposes specific metadata such as:
This would enable AI assistants to answer project-related questions such as:
Instead of project-specific customizations, this integration would leverage a standardized MCP server to securely allow AI assistants access. This server could exist as a separate integration that leverages Appwrite’s existing APIs and respects all existing authentication and permission systems. I’d be happy to implement this using Node.js if this concept aligns with Appwrite’s future development roadmap.
👀 Have you spent some time to check if this issue has been raised before?
🏢 Have you read the Code of Conduct?
All reactions