Evolving AI Agents

Darwin Godel Machine reference code for self-improving, self-evolving AI agents in Python

Works with setup

Test report

Verdict
Works with setup
Score
6.4/10
Tested
Jul 21, 2026
Environment
Claude Code 2.x (agent harness)

This skill is no longer available upstream. Our re-check on Aug 10, 2026 couldn't find it any more (SKILL.md not found in repo tree). The test below is what we measured on Jul 21, 2026 and we're leaving it up as a record — but there is nothing left to install, so we've removed the command.

Fetched the single self-contained SKILL.md at skills/ai-ml/self-evolving-agent (identical fork copy at skills/agent/self-evolving-agent-plurigrid-asi); metadata.json present, no external scripts referenced, no security smells. I ran two solvers on the same toy task (evolve a random 12-char string toward a target). My baseline elitist evolutionary loop reached fitness 1.0 in 85 generations; the skill's verbatim DarwinGodelMachine stalled at 0.667 after 2000 generations and never solved it. Observed its is_novel/code_similarity uses whitespace tokenization, so the diversity check is inert for non-multi-token candidates and the archive fills to 100 with low-fitness junk, drowning selection.

Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works

  • Installs cleanly 5/5
  • Triggers reliably 5/5
  • Output vs. baseline 3/10
  • Docs & honesty 3/5

What Evolving AI Agents does

Provides self-contained Python reference implementations of Darwin Godel Machine patterns for self-evolving AI agents: an agent archive, fitness-proportionate sampling, LLM-based code mutation, benchmark evaluation, novelty-based archiving, and long-term memory. Triggers when a user wants to build self-improving agents, evolutionary agent loops, or DGM-style code mutation and evaluation. It is a conceptual reference rather than a plug-and-run library; the stock evolution loop needs adaptation to converge on a given task.

How to install Evolving AI Agents

Nothing to install: the source repository no longer has this skill. If the author brings it back, our daily re-check will pick it up and the command will reappear here.

Commands — how to trigger Evolving AI Agents

  • /evolving-ai-agents Darwin Godel Machine reference code for self-improving, self-evolving AI agents in Python

It also activates on plain-language prompts like these:

  • Evolve this agent's prompt against our benchmark
  • Optimize my agent's skills automatically
  • Run an evaluation loop to improve this agent

Frequently asked questions

Is the Evolving AI Agents skill free?
Yes. The skill itself is free from majiayu000/claude-skill-registry. SkillProof publishes the install command and an independent test verdict at no cost.
Does Evolving AI Agents work with Claude Code?
We tested it with Claude Code 2.x (agent harness) on Jul 21, 2026. Verdict: Works with setup. Fetched the single self-contained SKILL.md at skills/ai-ml/self-evolving-agent (identical fork copy at skills/agent/self-evolving-agent-plurigrid-asi); metadata.json present, no external scripts referenced, no security smells. I ran two solvers on the same toy task (evolve a random 12-char string toward a target). My baseline elitist evolutionary loop reached fitness 1.0 in 85 generations; the skill's verbatim DarwinGodelMachine stalled at 0.667 after 2000 generations and never solved it. Observed its is_novel/code_similarity uses whitespace tokenization, so the diversity check is inert for non-multi-token candidates and the archive fills to 100 with low-fitness junk, drowning selection.
What is the Evolving AI Agents SkillProof Score?
6.4/10 — installs cleanly 5/5, triggers reliably 5/5, output vs. baseline 3/10, docs & honesty 3/5.
How do I install Evolving AI Agents?
Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
Can I use Evolving AI Agents with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.