#self-improving
GitHub repositories that have self-applied the topic "self-improving" — a creator-tagged metadata that surfaces how AI projects describe themselves.
REPOS Repos for #self-improving (top 12 by stars)
Agent skills that don't rot: research prior art, build and test skills on any agent, catch drift, and turn lessons from real use into checked fixes. Claude Code, Codex, Gemini CLI, Cursor, Copilot, OpenCode.
ArttuAn/agent-harness-skillsIDE skills for scaffolding AI agent harnesses: ReAct, plan-and-execute, self-evolving, reflexion, multi-agent, code agent, RAG memory
bitranox/bitranox-skillsClaude Code plugin marketplace: the bitranox skill collection (82 skills) plus a self-improve Stop hook. Skills invoke as /bitranox:<skill>.
moriwo-dev-ai/ama-teras欲しい機能を言うと、アプリが自分で作る。ただし暴走はしない。安全な自己進化型デスクトップAIエージェント / Self-evolving desktop AI agent with safety gates
jerry729/evocordSelf-hosted Discord bot on Claude Code channels: web search, voice messages, sandboxed code execution, and self-evolution with admin review.
CryptoGnome/khanAutonomous AI agent company in Rust - a CEO agent that hires and manages a team of LLM agents, delegates work in parallel, evolves its own prompts, tools and skills, and streams a live web activity log. Watch one running live at khanbot.fun. Deploys to Railway in minutes; works with any OpenAI-compatible API.
Walliiee/agent-harnessA drop-in self-improving memory + operations harness for AI agents (OpenClaw / Hermes / Claude Code / Codex). Layered memory, autonomous drift-fixing loop, evals, and one-command DR. Templates + scripts, no secrets.
ofirozon/self-improving-channel-agent tejaskembalkar-ship-it/DaedalusSelf-improving AI agent runtime: durable memory, skill creation from real workflows, scheduled automations, parallel subagents, and model-agnostic execution.
dverdysh/freepalpMulti-agent AI orchestrator that runs on free models — self-improving, MCP, token streaming, DAG, WebUI
akaalias/ballparkAutonomous Research with AI: can a UAV learn a city by heart? An AI agent running an autoresearch loop (81 experiments, $364) found a 3.1 MB neural memory that geolocates in-flight pictures of Berlin to ~27 m — no GPS, no map on board, built for a <$5 ESP32-P4 flight computer.
PetrouilFan/canaryCanary — a minimal, modular, self-improving agent harness with atomic releases, shared memory and an OpenAI-compatible API
RELATED Other topics · full topics ranking →
#claude-code
3,787#ai-agents
3,201#llm
2,090#mcp
1,866#claude
1,854#python
1,678#developer-tools
1,530#codex
1,376#ai
1,181Aggregated by case-insensitive match against topics_json of each repo's latest content snapshot. methodology