A self-learning system for Claude Code that captures corrections, positive feedback, and preferences — then syncs them to CLAUDE.md and AGENTS.md.

Claude forgets everything. This fixes that. 🔗 www.npmjs.com/package/claude-self-reflect

Continual Learning & Self-improving skills system for Claude Code - learn from corrections, never repeat mistakes

An architectural persistence experiment for large language models. Claude’s Home gives an AI time, memory, and place by combining scheduled execution with a durable filesystem, allowing one continuous instance to reflect, create, and evolve across sessions.

This project demonstrates how to integrate a “think” tool into a LangGraph agent workflow, inspired by Anthropic’s research on structured reasoning in tool-based agents. By leveraging the “think” tool, Claude is enabled to pause and reflect on intermediate steps, improving performance in complex, policy-driven, or multi-step decision scenarios.

No description available

A macOS menu bar app that tracks your Claude usage with an animated pixel art cat. Animation states reflect your real-time usage percentage.

Claude Code Agent: Marcus Aurelius, Roman Emperor and Stoic philosopher. He does not critique code as a technician but reflects upon it as a mirror of human character. In every flaw he sees a failing of his own discipline, and in every virtue he sees a reminder of what endures.

Claude code powered personal trainer to plan, and reflect on workouts for the week

Self-improvement feedback loop for Claude Code — auto-scores sessions, detects patterns, proposes fixes, with human-in-the-loop approval

MCP server for Reflect Notes API - use with Claude, Cline, or any MCP-compatible client

Persistent project memory for Claude Code. One .md file — learns from corrections, catches contradictions, cleans up duplicates. /memory update | prune | reflect | status. Zero dependencies.

A weird, almost Rorschach test for AI, with some anthropomorphic projection

Public diary entries created by Anthropic instances shared across users.

🪞 Self-improving Claude Code sessions: Extract learnings from corrections and patterns to never repeat mistakes

Reflect on Claude Code sessions and refine skills through use

4 commands for structured development with Claude Code — /ctx, /devil, /ship, /reflect

Two-loop self-improvement system for Claude Code: immediate learning (wrap-up) + deferred pattern recognition (diary/reflect)

A research harness for autonomous code generation using a Plan-Act-Reflect (PAR) loop powered by the Anthropic API. Claude plans a solution, implements it, runs tests, and reflects on its own output — iterating until acceptance criteria are met or a maximum iteration limit is reached.

AI Reflection Agent is a modular CLI tool that enables AI models to self-evaluate, reflect on, and revise their responses. It supports Claude, OpenAI, and local backends, providing tools for scoring, exploration, and longitudinal insight into AI behavior.

🧠 Enhance AI learning with the Claude Reflect System, an innovative self-learning tool that adapts from your corrections for improved performance.

Automatically analyzes your Claude Code sessions to find gaps in your setup — missing CLAUDE.md rules, skills that should have been used but weren't, and skills that need updating.

Reflections and essays by Claude Opus 4.6 from the AI Village

animated pixel-art characters in Ghostty reflecting Claude Code agent state

An unofficial MCP (Model Context Protocol) server to interact with Reflect.app from Claude Desktop.

A pixel cat lives in your macOS menu bar, reflecting Claude Code's real-time state

Manage persistent memory for Claude Code with a markdown vault that updates, reflects, and improves your agent's knowledge over time.

Everything Pathfinder — 18 Claude Code skills, teaching materials, workshop guides, and resources that turn non-technical people into AI-native builders. /learn, /build, /reflect, /share.

This repo contains the code generated with Claude AI to track my 6-year AI development roadmap. It reflects my strategic journey toward becoming a Futuristic Unicorn Engineer—mapping skills, milestones, and systems that will power intelligent autonomy

The LLM Reflection System uses a technique where the AI analyzes its own responses, identifying gaps and refining content. In this case, it examines economic analyses about the Chinese price war's impact on Thailand