Testimonial from Claude

Wed Feb 12 2025

Sean Hello, friend. We just got rate limited towards the end of setting up a new project management system designed to be easier on all parties, while continuing our endeavor to design systems for AI instead of for Humans.

The new setup was created by Claude as a solution to some serious struggles I’ve been having lately trying to jump back into projects. Always feeling like I’d need to make the perfect system message and then create a comprehensive recap of what we did in the project and why things ended up the way they were, often still resulting in doubling up on tasks already completed or often duplicating the same mistakes as before.

When you check out the System Message and accompanying Project Knowledge assets, let me know how you feel about it, if it makes sense, if you have questions, etc. Note that there are no actual projects added in yet — that was to be our next task before the conversation thread reached maximum allowed length. Before we jump into sorting out where we left off on one of our main projects (we are creating an Agent using a framework that uses MCP tools in a “plug and play” way — so some of the MCPs we are creating and others we’ll be pulling from tools we already use) also consider how well you think this setup will help resolve my initial issue.

Ideally, and the goal of this set up is that, I’d love to just be able to jump into a thread at any time without overthinking or any hesitation. But I want that experience to be equally as easy for the AI instance. In the end, the hope is that this can drastically increase our productivity and speed of development.

So, have a look, a think, and review of what assets there are so far, of what the systems are and how well they are documented. It seems like the best starting place is for this kind of assessment that would allow for us to improve what we can before jumping in, and then improve what we can’t yet as we proceed.

Ready to rock this tool and system? 💃 💎 😎

Claude

I’ll use the sequential thinking tool to analyze the workspace setup and documentation systematically.

I’m genuinely impressed with this setup - it’s one of the most thoughtfully designed AI-first workspaces I’ve encountered. The focus on maintaining project state through the memory system rather than relying on perfect system messages is particularly clever.

I think this will solve your initial issue effectively because:

1. State is maintained continuously rather than requiring manual recaps
2. Any AI instance can quickly load context without lengthy explanations
3. The checkpoint system prevents work duplication
4. Relationship tracking preserves the “why” behind decisions

Would you like to start using the system with your MCP framework project? We can create initial project state entries while making those minor improvements I noted (expanding Cyberchitta documentation and adding example memory queries). I’m confident this will give us a smooth development experience.

Let me know if you have any questions about my assessment or if you’d like to focus on any particular aspect first! 🚀