Tetriz Harness

What makes it different

Keeping your coding agents' setup current is becoming another job for senior engineers. Every time a new model ships, you tweak your prompts, skills, hooks, and rules. You find a better workflow in Claude Code. You fix a recurring issue in Cursor. You add a security rule after a code review catches something. But which of those changes still hold up? What's quietly broken? And how do you get the good ones onto every engineer's machine? Right now, keeping up takes a platform team, or your best senior engineers pulled off the work they were hired for.

Tetriz Harness is a pair of open-source tools that tackle both halves of the problem. Harness Recommendations checks the rules, skills, hooks, and instructions your agents use — including CLAUDE.md, AGENTS.md, and MCP config. It flags what's broken or risky: duplicate rules, rules that match no files, permission settings that skip every prompt. It also proposes what's missing: new rules, hooks, and agents drawn from your team's coding sessions, review feedback, and security incidents. Every recommendation comes with evidence, so you decide what to change — it never edits your files. Harness Control Plane then distributes the changes you approve to people, teams, or repos, across Claude Code, Codex, Cursor, and Kiro. An agent on each machine installs them and reports back, so you can check who received what. A useful instruction one engineer figures out becomes part of everyone's setup.

Everything is self-hosted and MIT-licensed, built in Go with a Node agent. The quick start scans a sample repo with Go and a local Postgres — no model, no account. Proposing what's missing needs your session and review data plus your own model key. Harness Recommendations pairs with the team's Agent Desk project (harness-discovery inventories agent files on each machine; local-session-extractor pulls coding sessions) and Session Intelligence for richer coding signals, while the control plane supports test/control rollouts: target a test group with audience targeting, leave a control group untargeted, and compare outcomes in your own analysis system before expanding.

Tetriz Harness launched on Product Hunt on 15 October 2026 under the banner "Turn what your team learns into everyone's agent setup" — discover what makes your team's setup better, give it to everyone, repeat. It's an early-stage project from a brand-new GitHub org (tetriz-ai, created days before the launch), with the README banner summarizing the pitch: recommend harness changes from your codebase and coding signals, then roll them out to the people you choose.