Ruflo

Best for: developers running several coding agents at once who need coordination, shared memory and guardrails around them.

Ruflo is an agent meta-harness for Claude Code and Codex. Its framing is that an agent equals a model plus a harness: the model writes, while the harness supplies tools, memory, loops, sandboxes and controls that let it do real work. Ruflo is that execution layer, adding over a hundred specialised agents, coordinated swarms, self-learning memory, federated communication across machines and security guardrails. Setup is npx ruflo init.

What makes it useful: coordination between agents rather than isolated runs. Agents self-organise into swarms and share memory, which is the practical difference between running several agents and having them work on the same problem coherently.

Where it falls short: this is a heavy abstraction over tools that already function on their own, and the complexity only pays off if your workflow genuinely needs multiple coordinated agents. Running many agents in parallel multiplies token costs quickly. Expect meaningful time learning the concepts before it becomes productive.

Details: TypeScript, MIT licensed. Installs via npx.