@Daniel Petro shared this with me the other day and I find it to be one of the most useful tools I have come across in a while for AI powered development: https://agentation.dev/ The core idea: instead of describing a bug or design issue to your AI agent ("the blue button in the top right looks off"), you literally point at it — and the tool gives your agent the exact information to find it directly in your codebase. Overview of how it works: 1. Click any element → write your note → done 2. Select text → annotate a typo or copy issue 3. Drag a box → annotate an entire area at once 4. Hit copy → paste structured markdown directly into Claude Code, Cursor, or any AI tool The output is structured: element path, React component hierarchy, classes, position, computed CSS, so your agent can search for it with accuracy. You can also connect it to an MCP server they built, so your agent sees annotations in real time. No copy-paste at all. Just annotate in the browser and your Claude Code session already has full context. There's also a "watch mode" where the agent loops continuously, picking up new annotations as you create them and resolving them after fixing. You annotate, it fixes, in real time. If you're building with Claude Code or Cursor and doing any kind of UI work, this fits right into that workflow. Worth 10 minutes to try. Thanks @Daniel Petro for the recommendation! 🙌

I used to restart Claude sessions all the time because I'd hit context limits or need to switch tasks. Every restart meant re-explaining my setup, tech stack, and what I was building. Then I started treating MCP servers like my 'external brain'. Instead of cramming everything into chat context, I built a simple MCP that reads my project docs, database schemas, and recent decisions from files. Now Claude just asks the MCP when it needs context. Now I can work on the same project across 10+ sessions without losing continuity. Claude remembers everything because the MCP remembers everything. Anyone else structuring their MCP servers this way? Would love to hear other patterns people are using.

Thanks to a great idea from Ray Deck, I have reduced the number of tools in the snappy.ai Xano MCP from ~180 tools and ~150,000 tokens to ~1000 tokens and 3 core tools: 1. Gemini Flash Powered Search - "What tools do I have?" (Semantic search across all tools - e.g., "create table", "list functions") 2. Info - "Tell me more about that specific tool" (returns parameters, examples, docs) 3. Execute - "Do it" (runs the tool with the right parameters) Now the initial handshake costs only ~1000 tokens instead of 150,000. Claude discovers tools as needed rather than loading everything upfront. This pattern should work for any MCP with dozens or hundreds of tools. Instead of exposing everything at once, create a search/discovery layer that lets Claude Code find what it needs when it needs it!

With the news of OpenAI leaning more into MCP - we now have THREE versions of the snappy.ai Xano MCP available for our users! mcp.snappysaas.com = uses the Snappy SDK to make it easier for AI Agents to build like an EXPERT in XanoScript, through a proprietary system of multi-layered autofixing and automatic AI review! raw.snappysaas.com = does NOT uses the Snappy SDK, for those who prefer to use raw XanoScript, which has its strengths in many cases! readonly.snappysaas.com = read only tools, for observability, and providing services such as a Slack Bot that can triage issues/answer client queries using it's advanced ability to gain insight automatically. Here is a changelog with some of the recent changes I have made in the last week: 🚀 New Features TypeScript Client Generator Tool (Beta Testing) - Automatically generate TypeScript clients for your Xano APIs - Type-safe interfaces with pagination support - Reduces integration time to frontend interfaces Enhanced Debugging System - Real-time debugging tips and context relevant suggestions - 80% faster error resolution by AI Agent - Interactive guidance for common XanoScript mistakes System Optimization - Reduction in response sizes (faster loading) - Removed 10 outdated tools for context performance - Fixed dynamic pagination issues XanoScript Capability - Implemented and enhanced function call syntax issues - Resolved Raw XanoScript API parameter formatting problems - Corrected "Unknown" function ID errors --- These updates are automatically available to all users!

Spent about 10 minutes on requirements and planning with Claude Code, kicked off the plan, took a shower, and came back to this... nice. (Snappy-Xano or V2 used on this one)