🦞 CRACKING THE CLAW - by Forest Mars - CTO Lunch NYC

kept by eddie feb 20

OpenClaw sacrifices the full observability of its minimal core (Pi) as it scales to a multi-agent gateway, creating un-auditable reasoning chains.

What I learned building an opinionated and minimal coding agent
The author of Pi shares design principles for a minimal coding agent: no background tasks, always YOLO mode, and full context control without vendor lock-in.
also on Mario Zechner, pi, OpenClaw, Anthropic, #open-source, #ai-coding, #developer-tools
Pi: The Minimal Agent Within OpenClaw
Pi, a minimal coding agent with a tiny core and extension system, shows how software can be built by agents for agents, pointing toward the future of development.
also on Mario Zechner, pi, OpenClaw, #ai-agents, #open-source, #ai-coding, #developer-tools
Import AI 455: Automating AI Research
The essay argues there is a 60%+ chance that no-human-involved AI R&D will occur by end of 2028, based on accelerating AI capabilities in coding, science, and engineering tasks.
also on DeepMind, Anthropic, #ai-agents
Import AI 441: My agents are working. Are yours?
AI agents can now perform complex research and coding tasks autonomously, dramatically multiplying human productivity, as experienced by Anthropic's Jack Clark.
also on DeepMind, Anthropic, #ai-agents
MCP is dead. Long live the CLI
The Model Context Protocol (MCP) offers no real benefit over command-line interfaces (CLIs), which are more composable, debuggable, and reliable for LLM tool use.
also on pi, OpenClaw, Anthropic, #ai-coding, #developer-tools