Let Skills Evolve Collectively with Agentic Evolver
WikiSkill (arXiv:2608.27454) for Hermes Agent — self-evolving agent skills via a persistent knowledge wiki. Faithful Algorithm 1 implementation with real agent runs, isolated skill gating, and a documented live run log.
Bounded, evidence-driven SKILL.md evolution under frozen evaluation and safety contracts.
DAG workflow engine for AI agents - DAG+FSM orchestration, pluggable step executors, three-tier memory (LRU/FTS5/semantic), automatic skill evolution. 74 tests, zero runtime deps.
Evidence-driven skill evolution for Hermes Agent — reports, dry-run proposals, candidate search, and guarded apply
Flow-driven recursive skill evolution for agentic LLM orchestration.
A curated collection of papers & repos on Auto-Skill (self-evolving agents) and Auto-Rubric (rubric learning from preferences) for LLM alignment & customization.
Pulse — Hermes-style self-improving AI agent. Reliability-first rebuild with evaluated skill self-evolution, multi-agent team orchestration, dialectic user modeling, and fully self-hosted default stack (Ollama + SQLite FTS5). Apache 2.0 license.
Multi-agent framework for Claude Code: tier-aware acceptance (S/M/L), acceptor/optimizer split, filesystem-isolated adversary review, cross-family second-opinion via Codex MCP, SkillOpt-style skill evolution, event ledger observability, human as supreme judge
SkillTV-Bench: a benchmark for evaluating, evolving, and analyzing judge skills on agent trajectories.
技能自进化系统 v2.0.0 — 8阶段Loop + 跨模型双裁判 + 30题测试集
WikiSkill: Compiling Agent Experience into Persistent Knowledge for Skill Evolution — for OpenCode, Claude Code, and Codex CLI
数字员工操作系统 — AI agent 像真正员工一样工作、学习、成长的协作平台。3D 虚拟办公室 + 技能进化 + 职业发展 + 记忆系统
Bionic memory for AI agents: GraphRAG, predictive memory, dream consolidation, governed skill evolution, MCP, REST API, and dashboard.
A control-theoretic learning plane for Hermes-style AI agents
Zero-dependency AI agent orchestration engine. Parallel DAG execution, skill genetics, ontology routing, human approval gates. 8.7K LOC, <1MB. The embeddable agent engine that learns.
Evolve native Codex + Claude Skills: discover, test, update, deduplicate, and remove—with bounded ReMe memory and reversible policy support.
MetaClaw fork: self-hosted GPU training (GRPO + LoRA + vLLM), per-agent skill isolation, multi-agent mode routing, training engine bug fixes. Train your AI agent on your own A100s. No cloud dependency.
Evolve AI agent skills through chat interactions that persist across devices and users with a collective learning framework.
Turn Codex work records into a personal wiki and reusable skill candidates, with source evidence and evaluation gates.
To associate your repository with the skill-evolution topic, visit your repo's landing page and select "manage topics."