# Best of Agent Harnesses and Harness Techniques

A ranked, weekly-updated catalog of AI agent harnesses and orchestration frameworks, with copy-paste templates and playbooks, and an MCP server and dataset an agent can query to pick one.

Best of Agent Harnesses and Harness Techniques is a curated, ranked list of AI agent harnesses, orchestration frameworks and harness techniques for building reliable agentic systems. A companion website lets visitors browse one page per harness and filter by capability, autonomy and recovery behavior.

The project distinguishes a model, which answers, from a harness, the surrounding runtime that decides what tools exist, what needs approval, what the model sees each turn, and what survives a crash; it ranks entries by relevance to those harness concerns and by activity, and rescores the list weekly. Alongside the ranking it publishes copy-paste templates, such as an AGENTS.md file and safe Claude Code settings, and step-by-step playbooks for setting harnesses up.

The list is also published as an MCP server with recommendation and pick-harness tools, and as an llms.txt and JSON dataset, so an agent can query it directly rather than a person reading the README.

Website: https://ryanalberts.github.io/best-of-Agent-Harnesses/