Continual agent skill evolution through persistent decision history. Whole-skill optimisation (SKILL.md + scripts + references) with every decision landing as a local Git issue / PR / wiki. Runs on any agentskills.io runtime — Claude Code, Codex, OpenClaw, Hermes.

Meta-skill for optimizing Claude Skill descriptions using SDS standard and data-driven analysis

Refactor any Claude Code skill by extracting deterministic operations into a Python script. Reduces SKILL.md size by 60-80% while preserving all AI judgment.

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

Measurement-disciplined optimizer for Claude Agent Skills — three-gate system (stability, effect size, function preservation), anchored rubric, train/holdout split. Inspired by Karpathy's autoresearch and alchaincyf's darwin-skill.

A production-grade, full-stack typing intelligence platform that operationalizes adaptive keystroke learning through dynamic content synthesis, real-time performance telemetry, and modular text generation pipelines—enabling precision skill acquisition, behavioral feedback loops, and scalable multi-language proficiency development.

Autonomous AI skill optimizer — inspired by Karpathy's autoresearch. Scores and auto-improves your agent SKILL.md files.

Eval-gated skill optimizer for coding agents — keeps only the skill edits that measurably raise your repo's real test pass rate. Rust CLI, no API key.

Simulation and analysis of employment skill optimization under AI-driven labor-market change.

Systematic methodology for optimizing agent skills using deep-learning-style approach. Based on arXiv:2605.23904

Train agent skills like neural networks — with epochs, batch size, learning rates, and validation gates — without touching model weights. Based on Microsoft SkillOpt.

Evidence-driven optimization and no-skill evaluation for Codex and Claude Skills; create small candidates only when real task evidence supports them.

a project that tries to combine the advantages of SkillVM and ArbiterOS and build a more reliable agent system.

Shouters Tips: How to Stay High Skill | A localized training utility designed to optimize competitive reaction loops and tactical positioning. It allows users to track performance analytics, configure customizable aim-trainer exercises, evaluate decision-making trees, and organize personal warm-up routines to maintain an elite mechanical baseline.

Train AI agent skills like model parameters. Offline text-space optimizer implementing SkillOpt (arXiv:2605.23904v2) for Hermes Agent. Bounded edits, validation gating, zero inference overhead.