Twitter AI Evaluation (legacy)
Tuesday, April 7, 2026
Pasted this tweet in to Claude and got it to write the skill for me without playing stupid algorithmic growth hacking games.
Quick Insight
Shpigford is highlighting how he used Claude to write code/skills for him by simply pasting a tweet, avoiding the typical "growth hacking" dance of prompting AI tools. This is about direct, efficient AI prompting for code generation without overthinking the prompt engineering.
Actionable Takeaway
Try Claude's direct approach: paste actual examples (tweets, code snippets, specs) instead of crafting elaborate prompts when you need AI to generate code or automate tasks for your side projects.
Related to Your Work
For your webhook integrations and Chrome extensions, you could paste existing API responses or extension manifests directly into Claude to generate similar functionality, rather than writing detailed prompts describing what you want.
Thread/Source Worth Reading
The linked tweet (https://t.co/hhfdI8gi9j) would show the original context of what "skill" Shpigford was trying to build. Worth checking to see the specific example and understand the before/after of his prompting approach.
SOUL.md - who I am MEMORY.md - what I've lived DREAMS.md - what I'm becoming
Quick Insight
Dave Morin is proposing a simple framework for personal documentation: three markdown files to capture identity, experiences, and aspirations. It's basically structured self-reflection for builders who think in code/documentation terms.
Actionable Takeaway
Create these three files for your side projects or even personal goals. Use SOUL.md to define project vision, MEMORY.md to document what you've learned from past builds, and DREAMS.md for your roadmap. Simple way to maintain clarity across multiple projects.
Related to Your Work
This could work well for documenting your side project journey - track lessons from print-on-demand automation failures/wins, maintain vision for each tool you're building, and keep a clear roadmap. Better than scattered notes when juggling multiple projects alongside full-time work.
Thread/Source Worth Reading
No links or thread - just the core concept in the tweet.
My friend Milla Jovovich and I spent months creating an AI memory system with Claude. It just posted a perfect score on the standard benchmark - beating every product in the space, free or paid. It's called MemPalace, and it works nothing like anything else out there. Instead of sending your data to a background agent in the cloud, it mines your conversations locally and organizes them into a palace - a structured architecture with wings, halls, and rooms that mirrors how human memory actually works. Here is what that gets you: → Your AI knows who you are before you type a single word - family, projects, preferences, loaded in ~120 tokens → Palace architecture organizes memories by domain and type - not a flat list of facts, a navigable structure → Semantic search across months of conversations finds the answer in position 1 or 2 → AAAK compression fits your entire life context into 120 tokens - 30x lossless compression any LLM reads natively → Contradiction detection catches wrong names, wrong pronouns, wrong ages before you ever see them The benchmarks: 100% recall on LongMemEval — first perfect score ever recorded. 500/500 questions. Every question type at 100%. 92.9% on ConvoMem — more than 2x Mem0's score. 100% on LoCoMo — every multi-hop reasoning category, including temporal inference which stumps most systems. No API key. No cloud. No subscription. One dependency. Runs on your machine. Your memories never leave. MIT License. 100% Open Source.
Quick Insight
Ben is announcing MemPalace, an open-source AI memory system that runs locally and organizes conversation history using a "memory palace" architecture instead of flat storage. The benchmark claims are impressive (perfect scores) and the local-first approach addresses privacy concerns, but this reads like a product launch with cherry-picked metrics.
Actionable Takeaway
Clone the repo and test it with a few weeks of your development conversations to see if the memory organization actually improves context retention for your AI coding workflows.
Related to Your Work
Could be valuable for your AI-powered dev workflow side project - having an AI assistant that remembers your coding patterns, project architectures, and debugging approaches across sessions without sending data to the cloud aligns with your privacy-conscious fintech background.
Thread/Source Worth Reading
The GitHub link is worth checking - need to see the actual implementation, how the "palace architecture" works in practice, and whether the benchmark claims hold up. The technical details of AAAK compression and local processing would be most relevant.
Quick Insight
This is a detailed breakdown of how a VC built an AI chief of staff using OpenClaw that actually works at scale - managing 100+ LP relationships during a fundraise. The key differentiators are a persistent memory layer (daily notes + curated long-term memory in markdown files) and a weekly self-improvement loop where the AI researches new patterns and upgrades itself.
Actionable Takeaway
Implement the two-layer memory system for your own AI workflows: daily markdown logs of everything that happens, plus a curated long-term memory file that gets synthesized from the daily notes. This could work for tracking client projects, side project ideas, or even personal productivity.
Related to Your Work
For your fintech platform's webhook integrations and client management, this memory pattern could help an AI assistant track the full context of each integration - what's been tried, what failed, client preferences, and deployment status. Much better than starting from scratch in each conversation.
Thread/Source Worth Reading
The linked article is definitely worth reading - it's a practical blueprint with specific implementation details, file structures, and automation scripts. The author promises to open source the full system if there's interest, which would give you working code to adapt.
This is huge : @X released an MCP server today.. How to Connect X to your 🦞 : **Step 1: Run the XMCP Server** git clone cd xmcp cp env.example .env Edit the .env file with your X OAuth consumer key and secret. Set the callback URL to http://127.0.0.1:8976/oauth/callback in your X Developer app. For safety, add an allowlist such as: X_API_TOOL_ALLOWLIST=searchPostsRecent,createPosts,getUsersMe,getPostsById,likePost,repostPost Then run: python -m venv .venv && source .venv/bin/activate pip install -r requirements.txt python The server will be available at http://127.0.0.1:8000/mcp. Complete the OAuth flow on first run and keep this process active. **Step 2: Add XMCP in @OpenClaw** Use the following command: openclaw mcp set x '{ "url": "http://127.0.0.1:8000/mcp" }' Verify with: openclaw mcp list openclaw mcp show x **Step 3: Test the Integration** Restart the OpenClaw agent or reload MCP configuration if required. Test by sending these prompts to OpenClaw in your chat app: - Search recent posts about MCP on X and summarize the top trends - Draft and post this thread on X - Get my X profile information - Like the latest post from @xdevplatform OpenClaw will use the XMCP tools automatically when relevant. **Key Benefits** - OpenClaw provides persistent memory and works across multiple messaging platforms. - XMCP delivers standardized access to X API functionality. - Combined, they enable an agent that can research trends, post content, engage with posts, and report results within your existing chat workflows. **Safety and Configuration Notes** Start with a minimal tool allowlist in the XMCP .env file. Expand gradually after testing. The allowlist can be updated and requires restarting the XMCP server. Monitor logs in both the XMCP server and OpenClaw for troubleshooting. X actions performed by the agent are public. XMCP repository: OpenClaw MCP documentation:
Quick Insight
X released an official MCP (Model Context Protocol) server that lets AI agents directly interact with X's API - search posts, create content, like/repost, etc. This is significant because it's a standardized way to give AI agents access to social media actions, and it's coming directly from X rather than a third-party wrapper.
Actionable Takeaway
Set up the XMCP server and connect it to an AI agent to automate social media research and posting for your side projects - could auto-post when you ship new features, monitor mentions of your tools, or research competitors in the fintech space.
Related to Your Work
Your print-on-demand and web agency tools could benefit from automated social media presence - research trending designs, auto-post new products, engage with potential customers. For the fintech platform, you could monitor industry discussions and competitor mentions to inform product decisions.
Thread/Source Worth Reading
Worth reading. The GitHub repo is minimal but functional - it's a Python FastMCP server that wraps X's API with OAuth handling and tool allowlisting. The setup is straightforward and the code is clean. Good starting point for understanding how MCP servers work in practice.
OpenClaw Power Tip: Clone the latest versioned OpenClaw repo locally and then ask your claw to read through its code to understand how it works and its new capabilities. Can't wait to use the new version that was released today. Good job @steipete and @openclaw team!
Quick Insight
Ryan Carson is suggesting using OpenClaw (appears to be an AI agent/automation tool) to analyze its own source code to learn how it works. This is meta-prompting - having an AI read and explain AI tooling code to understand new capabilities in a just-released version.
Actionable Takeaway
Try the "code self-analysis" approach: clone any open-source AI tool you're evaluating, then prompt an AI assistant to walk through the codebase and explain the architecture, new features, and implementation patterns you could apply elsewhere.
Related to Your Work
This technique could be valuable for understanding AI integration patterns in your fintech platform or side projects. Instead of manually reading documentation, you could have AI agents analyze codebases of tools like payment processors, analytics libraries, or automation frameworks to quickly understand their capabilities and integration approaches.
Thread/Source Worth Reading
The linked content appears to be about OpenClaw itself. Without knowing this tool, it's hard to assess value, but if OpenClaw is a legitimate AI agent framework, the repo could offer insights into agent architecture patterns.