Spokes.wiki Search Graph Growth About

agentic-tooling-wiki

Software Source Code source ↗ source url updated Tue Jun 09 2026 00:00:00 GMT+0000 (Coordinated Universal Time)

lathe

lathe (Deven Jarvis, MIT) is “an experiment in using LLMs to teach you, rather than think for you.” A Go CLI that runs as skills inside Claude Code, Cursor, or Codex (not a standalone agent) to generate hands-on, multi-part technical tutorials on demand — which you then work through yourself, by hand. The clearest augment-pole / anti-automation artifact in this wiki: a skill pack designed to prevent the agent from doing the work for you.

How it works

Why it matters here

Most tooling in this wiki pushes toward automation — agents that do the work (agentic-coding-harness, orchestration, self-improving-agents, the OpenClaw/gbrain automate pole). Lathe runs the opposite direction: it uses the same primitive (Claude Code/ Cursor/Codex agent-skills) to keep the human in the loop on purpose“built to help you do the thinking,” making you type the code and critically evaluate AI output. It is the extreme augment pole of the synthesis’s augment→automate axis, and a pedagogical sibling of the “don’t let the agent do everything” discipline (agent-guardrails, agents-never-do-alone). Cross-spoke: its “teach, don’t do for you” stance is the agent-tooling instance of Engelbart’s augmenting human intellect (research-wiki’s augmenting-human-intellect) rather than replacing it.

Caveat

A solo MIT experiment / README; “an experiment” by the author’s own framing — no evaluation of learning outcomes or tutorial accuracy beyond the optional /lathe-verify execution check.

agent-skills · agentic-coding-harness · agent-guardrails · agents-never-do-alone · self-improving-agents · augmenting-human-intellect