Harness Hub

Community · 13 playlists

Playlists

Learning paths put related videos in an order you can follow: tutorials first, then deep dives, then talks. Binge lists are the freshest and most-watched videos, rebuilt every night. Press play once and the whole list runs; tick off what you have watched to pick up where you left off. All from the video library.

Binge-watch

Binge-watch: this week

Every long-form video from the last seven days, newest first, on any topic. Press play and let it run.

Most watched, last two weeks

The recent videos people are actually watching, ranked by YouTube views and refreshed every night.

Conference talks marathon

Talks from the last two months of harness engineering conferences and meetups, newest first.

Talks & interviews

Learning paths

Harness engineering, the fundamentals

What a harness is, why it matters as much as the model, and how teams design the loop, tools and context around an agent.

Harness engineering

Orchestrating many agents

Parallel agents, worktrees, software factories and the orchestration patterns behind them, from first experiments to production setups.

Multi-agent orchestration

Context and memory

How agents keep what matters: context engineering, memory files, retrieval and compaction, in the order you would learn them.

Context engineering & memory

MCP, tools and skills

Give agents hands: MCP servers, tool calling, agent skills, plugins and slash commands, from basics to building your own.

MCP & tools · Skills, plugins & mods

Codex and OpenAI's agents

The Codex CLI and app and OpenAI's agent stack, from getting started to the workflows heavy users rely on.

Codex & OpenAI agents

Cursor and the IDE agents

Cursor, Copilot, Antigravity, Kiro and the other editor and terminal agents: how to use them and how they compare.

Cursor & IDE agents

Open harnesses and personal agents

opencode, Pi, Gemini CLI and building your own harness, plus always-on personal agents like OpenClaw and Hermes.

Open harnesses · OpenClaw, Hermes & personal agents

Shipping agents safely: security, cost and evals

What to sort out before agents run unattended: sandboxing and prompt injection, cost and rate limits, and measuring whether it works.

Security & sandboxes · Cost & limits · Evals & benchmarks

The Harness Hub newsletter

New harness companies, projects, releases and the monthly State of Harness report, in your inbox.