What 40+ coding-agent harnesses do between your prompt and the model, dissected
blaizedsouza · x · 2026-11-11
sermakarevich published part 1 of "Same model, different agent": a systematic overview of 40+ harnesses — the programs wrapped around the model. A harness takes your prompt, adds a system prompt, instruction files and a tool list, sends it to the model, and runs the tools the model requests.
The article compares harnesses across the loop structure, first-request construction, tools, subagents, output handling and permission mechanisms — and what those differences mean for users, explaining why the same model behaves drastically differently across agent products.
More from coding & agent
- Stop burning tokens: a Claude Code orchestrator prompt that routes tasks to cheaper models — TawohAwa · 2026-11-11
- Amazon paper: multi-agent setups burn 1.4-4.7x tokens for small gains; single agent is the default — rohanpaul_ai · 2026-11-11
- Dev shows off building a retro FPS entirely via Codex Mobile — 503 tests passing — Dimillian · 2026-11-11
- Explosion launches Ellf directory: 52 NLP templates plus multi-agent annotation workflow — _inesmontani · 2026-11-11
- Meme: 'Stochastic AI Agent Architecture' Is Just a Markov Chain in Disguise — PrasoonPratham · 2026-11-11
- OpenCode adds Gemini 4 Carbon and Claude Fable 5.2 support — dean_rie · 2026-11-11