Field notes from storage internals to AI dev tooling — wire protocols, write paths, and the methodologies that keep production code honest.
An interactive protocol explorer tracing every step of a write operation — from userspace syscall through kernel redirectors, authentication, cache management, distributed replication, and the critical ACK decision point. Compare SMB and NFS side by side.
A data-driven approach to prioritizing test coverage across 13 data properties and 16 test types. Interactive D3 heatmap reveals where testing effort yields the highest return.
A comprehensive deep-dive into modern browser automation — architecture, AI/MCP integration, locator systems, and how Playwright stacks up against Selenium and Cypress.
Most AI-for-code tools improve context by sending more of your code to a third party. Draft’s graph engine does the opposite — a local, deterministic, version-controlled knowledge graph. No MCP, no SaaS, no embeddings, no leaked code.
Amazon’s Kiro and Draft both solve the same problem: AI coding without structure is chaos. The honest comparison — bug hunting, ACID enforcement, three-stage review, architecture discovery, IDE portability, and a $0 price tag.
Before /draft:implement writes a line, Draft’s graph engine answers the question your AI assistant cannot: 47 downstream files, 12 tests, 3 config templates — and a refactor plan reshaped before any code lands. Three loops, one shared graph.
/draft:decompose WalkthroughA real run on a 14-file track. Eight modules, a circular dependency caught and broken before any code was written, and an LLD with contracts for the two high-complexity modules. Decomposition isn’t a planner — it’s an architectural guardrail.
Loading 200 MCP tool schemas upfront costs 50–140K tokens, breaks selection accuracy past ~30–50 tools, and slams into a 128-tool API ceiling on two of the four major providers. A comparative read on Claude Code, Cursor, Codex CLI, and GitHub Copilot as of May 2026.