What This AI Meetup Needs is a Webpage

Vibe coding w Bitcoin Maxi’s via Socratic #3 topics

• Agents
• OpenAI Codex: https://openai.com/index/introducing-codex/

• Google Coding Agent: https://jules.google
• Github Copilot https://github.blog/changelog/2025-05-19-github-copilot-coding-agent-in-public-preview/
• Claude Code SDK: https://docs.anthropic.com/en/docs/claude-code/sdk
• https://github.com/OpenSecretCloud/Maple/issues/120
• Amp https://ampcode.com/how-i-use-amp
• Don’t build multi agents: https://cognition.ai/blog/dont-build-multi-agents
• Anthropic agent research system: https://www.anthropic.com/engineering/built-multi-agent-research-system

• LLM memory: https://grantslatton.com/llm-memory
• Circuit tracing tools for LLMs https://www.anthropic.com/research/open-source-circuit-tracing
• Jack vibing https://github.com/block/goose/issues?q=is%3Apr+is%3Aopen+author%3Ajackjackbits
• AI Skeptic: https://fly.io/blog/youre-all-nuts/
• Models
• CPU music model https://x.com/StabilityAI/status/1922675163411497094
• Devstral dev model: https://mistral.ai/news/devstral
• New model for windsurf: https://windsurf.com/blog/windsurf-wave-9-swe-1
• Apple https://machinelearning.apple.com/research/apple-foundation-models-2025-updates
• Claude 4 
• https://www.anthropic.com/news/claude-4
• https://x.com/benhylak/status/1925619639859478644
• https://simonwillison.net/2025/May/25/claude-4-system-prompt/
• Gemini 2.5 Family: https://blog.google/products/gemini/gemini-2-5-model-family-expands/
• Creative Commons model: https://huggingface.co/blog/stellaathena/common-pile
• Magistra (Mistral Thinking Model)l: https://mistral.ai/news/magistral
• Psyche
• https://nousresearch.com/nous-psyche/
• https://psyche.network/
• Evals
• Eval startups fail: https://thomasliao.com/eval-startups
• Privacy
• Signal prez https://x.com/mer__edith/status/1925236198672314615
• Secure minions https://ollama.com/blog/secureminions
• MCP
• GitHub MCP vulnerability: https://invariantlabs.ai/blog/mcp-github-vulnerability
• Blocks Playbook for MCP: https://engineering.block.xyz/blog/blocks-playbook-for-designing-mcp-servers
• ChatGPT adds MCP: https://platform.openai.com/docs/mcp
• Research
• RLVR on Qwen2.5-Math models with completely random or incorrect rewards: https://rethink-rlvr.notion.site/Spurious-Rewards-Rethinking-Training-Signals-in-RLVR-1f4df34dac1880948858f95aeb88872f
• Apple paper about reasoning llms https://x.com/RubenHssd/status/1931389580105925115
• Claude responds to Apple’s Paper: https://www.alphaxiv.org/abs/2506.09250
• ChatGPT is making us stupid: https://www.brainonllm.com/

Socratic #3 Ideas

• https://x.com/xai/status/1923183620606619649
• Cool podcast about BAML – typesafe DSL for language model interactions. Guy uses small language models instead of LLMs, and makes them expert at narrow tasks. Need for better “ontologies” for multi-modal graph RAG e.g. Lettria. Big fan of Kuzu.
• https://open.spotify.com/episode/3agvlypsGrXZa0Hkw4m1wc
• New model for windsurf: https://windsurf.com/blog/windsurf-wave-9-swe-1
• Psyche https://x.com/NousResearch/status/1922744483571171605
• CPU music gen model https://x.com/StabilityAI/status/1922675163411497094
• OpenAI Codex: https://openai.com/index/introducing-codex/
• LLM memory: https://grantslatton.com/llm-memory
• Google Coding Agent: https://jules.google
• GitHub Coding Agent: https://github.blog/changelog/2025-05-19-github-copilot-coding-agent-in-public-preview/
• mobile first gemma model https://developers.googleblog.com/en/introducing-gemma-3n/
• Devstral dev model: https://mistral.ai/news/devstral
• https://x.com/mer__edith/status/1925236198672314615
• Eval startups fail: https://thomasliao.com/eval-startups
• OpenAI adds MCP: https://openai.com/index/new-tools-and-features-in-the-responses-api/
• Jony Ive startup https://www.bloomberg.com/news/articles/2025-05-21/openai-to-buy-apple-veteran-jony-ive-s-ai-device-startup-in-6-5-billion-deal
• https://x.com/bengeskin/status/1925552927885640124
• Reason-ModernCOLBERT reasoning-intensive retrieval model https://x.com/antoine_chaffin/status/1925555110521798925
• Claude 4 https://www.anthropic.com/news/claude-4
• https://x.com/benhylak/status/1925619639859478644
• https://simonwillison.net/2025/May/25/claude-4-system-prompt/
• Gemini diffusion https://simonwillison.net/2025/May/21/gemini-diffusion/
• The way of code: https://www.thewayofcode.com
• Finding 0days in the Linux kernel: https://sean.heelan.io/2025/05/22/how-i-used-o3-to-find-cve-2025-37899-a-remote-zeroday-vulnerability-in-the-linux-kernels-smb-implementation/
• No rag for coding agents: https://pashpashpash.substack.com/p/why-i-no-longer-recommend-rag-for
• GitHub MCP vulnerability: https://invariantlabs.ai/blog/mcp-github-vulnerability
• Mistral agents api https://x.com/MistralAI/status/1927364741162307702
• Relace coding models https://news.ycombinator.com/item?id=44108206
• https://x.com/ctnzr/status/1927391895879074047
• RLVR on Qwen2.5-Math models with completely random or incorrect rewards: https://rethink-rlvr.notion.site/Spurious-Rewards-Rethinking-Training-Signals-in-RLVR-1f4df34dac1880948858f95aeb88872f
• Circuit tracing tools for LLMs https://www.anthropic.com/research/open-source-circuit-tracing
• Human in the loop for MCP: https://modelcontextprotocol.io/specification/draft/client/elicitation
• AI diplomacy https://x.com/alxai_/status/1928504013030347180
• Claude  code analysis: https://southbridge-research.notion.site/claude-code-an-agentic-cleanroom-analysis
• https://fly.io/blog/youre-all-nuts/
• Guide to meta prompting: https://www.prompthub.us/blog/a-complete-guide-to-meta-prompting

• Secure minions https://ollama.com/blog/secureminions
• New Gemini 2.5 Pro https://blog.google/products/gemini/gemini-2-5-pro-latest-preview/
• Creative Commons model: https://huggingface.co/blog/stellaathena/common-pile
• Apple paper about llms https://x.com/RubenHssd/status/1931389580105925115
• https://x.com/jsngr/status/1932200732994019641
• Magistra (Mistral Thinking Model)l: https://mistral.ai/news/magistral
• Don’t build multi agents: https://cognition.ai/blog/dont-build-multi-agents
• Claude responds to Apple’s Paper: https://www.alphaxiv.org/abs/2506.09250
• Anthropic agent research system: https://www.anthropic.com/engineering/built-multi-agent-research-system
• ALE Bench: https://sakana.ai/ale-bench/
• Blocks Playbook for MCP: https://engineering.block.xyz/blog/blocks-playbook-for-designing-mcp-servers
• Monitoring sabotage: https://www.anthropic.com/research/shade-arena-sabotage-monitoring
• OpenAI preventing misalignment: https://openai.com/index/emergent-misalignment/

Eval Workshop

• Why evals
• xAI “unauthorized system prompt changes”
• Experience with OpenAI Evals
• OpenAI API tutorial
• https://platform.openai.com/evaluations
• Nice tutorial 
• Promptfoo
• Tweet generation
• Movies: external data in csv, compare models, js evaluation
• Jailbreak
• Huggingface Rag Evaluation 
• https://huggingface.co/learn/cookbook/en/rag_evaluation
• Ragas