Writing

We rent the model and inherit the harness, but we build the environment. Part II on deterministic gates, evals for skills, and the economics of the agent factory.

AI makes finding possible vulnerabilities cheap. The scarce resource is proving which ones are real. How we trace attack paths across code, infrastructure and authority boundaries.

When writing code stops being the bottleneck, review and correctness take its place. How we're promoting AI agents to first-class members of the team.

Every agent harness wants its own config format. We built a vendor-agnostic template so skills, rules and MCP servers are defined once and run everywhere.

What started as 'generate 10 meme images' turned into a case study in validation, iteration, and why semantic AI evaluation beats pixel heuristics. A 70% false positive rate forced us to rethink.

The journey from AI Skeptic to AI Native follows the exact same map as the Bitcoin journey. If you found your way to 'tick tock next block,' you already have the playbook.

Channel liquidity management is Lightning's hidden tax. PeerSwaps changes the economics entirely.

Moving a high-volume production node isn't copy-paste. It's open-heart surgery with uptime requirements.

We built an interactive playground for developers to experiment with BOLT12. Here's why developer experience matters.

BOLT12 represents a fundamental upgrade to Lightning's payment UX. Here's the engineering breakdown.

Fee estimation is deceptively hard. Here's how we blend multiple signals for reliable transaction confirmation.

A year of building Zap. The technical decisions, the community, and what I learned shipping open source.

My first steps into Lightning development. Building Zap wallet from zero knowledge to production.