Projects
05
mcp-dyno
An open-source CLI that measures how an MCP server performs when an LLM drives it — across efficiency, cost, context-bloat, correctness, and reliability — and proves before/after changes with paired statistics. It grew out of benchmarking dtc-mcp, when measurement turned out to be harder than the optimization.
Lenny's Round Table
A virtual round table built on 4 years of Lenny's Podcast. Drag three guests onto a felt, ask a question, and they respond as themselves — grounded in their real transcripts, with verbatim-validated quotes. They address each other. Sometimes they disagree. Sometimes only one of three has anything to say and the other two stay quiet.
dtc-mcp
A code-execution MCP server for Klaviyo + Shopify analytics. Three tools instead of 28, a stateful V8 sandbox so iterative analyses don't re-fetch, and a benchmark against the official Klaviyo MCP that drove the architecture.
A Penny for Your Thoughts
An interactive view of Anthropic's Natural Language Autoencoders. Ask an open-weight model a question, then spend a virtual penny to see its layer-32 activations translated back into language.
Proto-Review
A pinned-comment, scored, and screen-recorded review tool for Claude Design prototypes. Upload a zip, share a link, watch the replay.