15 Terms Every AI Product Builder Should Know, From Harness to Hooks
aakashgupta · x · 2026-10-12
Aakash Gupta maps the emerging vocabulary of AI product building:
- Harness: everything around the model that lets it act, remember, and stop. Anthropic moved the same model's Terminal-Bench 2.0 score by up to 6 points by changing only the harness — more than the gap between top leaderboard models.
- Context layer: Instructions File (read every task, e.g. CLAUDE.md), Rules (standing instructions as files), Context Library (the folder the agent reads first).
- Action layer: Tools (MCP, apps, terminal), Skills (playbooks loaded on demand), Subagents (helpers with fresh context).
- Safety layer: Permissions, Sandbox, Human-in-the-Loop — citing a dev whose home directory was wiped by rm -rf after a sandbox failed.
- Quality layer: Checks (evals, linters, second-model review) and Hooks (rules enforced by code the model can't skip).
Takeaway: Agent = Model + Harness; a decent model with a great harness consistently beats a great model with a bad one.
Related event: A Visual Guide to 15 Key Terms for Building AI Products(2 posts)→
More from coding & agent
- LangChain Founder: Persistent Machines Are a Weakness — Agents Run Better Fully Parallel — zeeg · 2026-10-12
- MCP servers eat 29k tokens before you type a word — dev ships mcp-tax CLI to measure the tax — princejain756 · 2026-10-12
- Opus 5.5 builds its own SVG drawing tool, then drafts a detailed PS5 controller cutaway — ImportGuy · 2026-10-12
- Cross-harness agent messaging: one Redditor's file-based bus spiraled into thousands of messages — positiveconstraint · 2026-10-12
- GPT-6 Astra nails 72-hour agent tasks with 63% on-time rate, crushing Fable 5.1's 4% — maksym_andr · 2026-10-12
- Voice-controlling your coding agent from iOS and resuming sessions at your desk: the always-on Mac setup debate — TheZachMueller · 2026-10-12