Darwin Mode

Darwin Mode is an AI Engineering skill for Claude Code, published by ruvnet in metaharness.

609 stars72 forkson ruvnet/metaharnessAdded 2026/08/25+2% in starsRepository updated 2026/08/24

agent-harnessagent-harness-generatoragent-scaffoldingagentic-aiagentic-frameworkautonomous-agentsclaude-codecodexcreate-agent-harnesshermes-agentmcpmcp-servermulti-agentopenai-codexopenclawpi-devrvmsandboxingscaffoldscaffolding

Install in seconds

Install Darwin Mode

Copy Darwin Mode into your Claude Code skills folder. Run the command in your terminal, or review the source on GitHub before installing.

terminal

npx degit https://github.com/ruvnet/metaharness/tree/main/kimi-k3-harness/.claude/skills/evolve ~/.claude/skills/evolve

Requires Node.js. Downloads this skill only — not the rest of the repository — into your Claude Code skills folder.

Without Node.js

git clone https://github.com/ruvnet/metaharness.git

Clones the whole repository, then copy the skill’s own directory into your skills folder yourself.

In this catalog

Source file: kimi-k3-harness/.claude/skills/evolve/SKILL.md in ruvnet/metaharness
Installs to: ~/.claude/skills/evolve
Collection: One of 16 skills cataloged from this repository
Category: AI Engineering — 2451 skills

What Darwin Mode does

Darwin Mode evolves a frozen-model harness by mutating one surface file per generation, sandboxing each child, and keeping only variants that measurably improve. Use it to run controlled self-improvement experiments on agent harness code.

Darwin Mode is cataloged under AI Engineering on DirSkills. Darwin Mode comes from a repository tagged agent-harness, agent-harness-generator, agent-scaffolding, agentic-ai and agentic-framework.

Documentation

README

Excerpt

evolve — Darwin Mode self-improvement

kimi-k3-harness ships with Darwin Mode (@metaharness/darwin, ADR-070…146): the model is frozen; the harness evolves. Each generation mutates ONE of the 7 surface files (planner, contextBuilder, reviewer, retry/tool/memory/score policy), sandboxes each child, scores it, and keeps only variants that measurably improve — building an archive of successful descendants.

Run it

npm run evolve        # real substrate: runs your test command per variant (deterministic mutator — no API key, no network)
npm run evolve:dry    # mock substrate: fast, fully offline, no test execution

Or directly:

npx metaharness-darwin evolve . --sandbox real --generations 3 --children 4

This is the opening of the README. Read the full README on GitHub.

Frequently asked about Darwin Mode

What else does ruvnet publish alongside Darwin Mode?
Darwin Mode is one of 16 skills that DirSkills catalogs from ruvnet/metaharness, the repository it ships in. Its siblings there include Compare Harnesses, Create Harness and Diag Harness. Each one is a separate skill with its own page in this directory, installs the same way Darwin Mode does, and is maintained by ruvnet in that same repository. The rest of the collection is listed on the ruvnet/metaharness page.
How does Darwin Mode compare to other AI Engineering skills?
Darwin Mode ranks #1925 by stars among the 2451 AI Engineering skills in this catalog. The most-starred ones next to it are Architecture Decision Records, AI-First Engineering and Agentic OS. DirSkills ranks by the star count of the repository each skill ships in, so that order reflects how popular those repositories are rather than any review of Darwin Mode against them. Open each page to compare what they document and how they install.