https://evanoman.com Software engineering, applied math, and AI en-us Sun, 19 Jul 2026 03:22:37 GMT https://evanoman.com/blog/2026/readwise-plus-mcp https://evanoman.com/blog/2026/readwise-plus-mcp Thu, 16 Jul 2026 00:00:00 GMT readwise-plus, my homegrown Python SDK for the Readwise API, ships an MCP server as of 0.3.0. pip install readwise-plus[mcp] and your agent can search, save, and organize your Readwise. https://evanoman.com/blog/2026/deep-research-workflows https://evanoman.com/blog/2026/deep-research-workflows Fri, 03 Jul 2026 00:00:00 GMT How a 274-line prose-orchestrated research skill became a short script, and what Claude Code's dynamic workflows actually do. https://evanoman.com/blog/2026/model-routing-mid-2026 https://evanoman.com/blog/2026/model-routing-mid-2026 Thu, 02 Jul 2026 00:00:00 GMT Research synthesis on cost levers for agentic coding in mid-2026, finding that caching and effort tuning outperform learned routing, and cheaper tokens do not mean cheaper tasks. https://evanoman.com/blog/2026/building-this-site https://evanoman.com/blog/2026/building-this-site Wed, 10 Jun 2026 00:00:00 GMT A Lovable scaffold, a 9,800-line refactor, four parallel agents, and the bugs that browser validation caught. https://evanoman.com/blog/2026/voyager-1 https://evanoman.com/blog/2026/voyager-1 Mon, 25 May 2026 00:00:00 GMT Real JPL ephemeris data and a handful of charts that answered the questions I actually had about Voyager 1's speed, distance, and trajectory. https://evanoman.com/blog/2026/critique-vs-absorption https://evanoman.com/blog/2026/critique-vs-absorption Mon, 18 May 2026 00:00:00 GMT A short thought on two reading styles — one absorptive, one adversarial — and what the research says about where each one breaks. https://evanoman.com/blog/2026/contour-integral https://evanoman.com/blog/2026/contour-integral Thu, 14 May 2026 00:00:00 GMT Dependency-free Python and a small C kernel for adaptive contour integration. The interesting half is the path. https://evanoman.com/blog/2026/simplex-from-scratch https://evanoman.com/blog/2026/simplex-from-scratch Thu, 14 May 2026 00:00:00 GMT Wrote a two-phase primal simplex in pure Python to see what Claude could do. https://evanoman.com/blog/2026/context-cap https://evanoman.com/blog/2026/context-cap Wed, 29 Apr 2026 00:00:00 GMT A portable analyzer that reads your Claude Code session shape and recommends a CLAUDE_CODE_AUTO_COMPACT_WINDOW. https://evanoman.com/blog/2026/agents-grading-the-cli https://evanoman.com/blog/2026/agents-grading-the-cli Tue, 28 Apr 2026 00:00:00 GMT Twelve sub-agents, no source access, one --help flag. A regression eval for agent ergonomics. https://evanoman.com/blog/2026/agentcore-buildout https://evanoman.com/blog/2026/agentcore-buildout Sun, 26 Apr 2026 00:00:00 GMT A small data-analysis agent on AWS Bedrock AgentCore Runtime. Scaffold in progress. https://evanoman.com/blog/2026/anthropic-challenge https://evanoman.com/blog/2026/anthropic-challenge Sun, 25 Jan 2026 00:00:00 GMT Anthropic's open-sourced performance take-home. Baseline 147,734 cycles, final 1,269 — 116× speedup, past every published model benchmark.