Reading
A complete Claude Code plugin system that enables the construction, analysis, and maintenance of agent systems through 21 specialized skills and 5 review agents, covering the full lifecycle from project initialization to continuous quality assurance.
Section 01
A complete Claude Code plugin system that enables the construction, analysis, and maintenance of agent systems through 21 specialized skills and 5 review agents, covering the full lifecycle from project initialization to continuous quality assurance.
Section 02
As AI coding assistants evolve from simple code completion tools to agent systems capable of performing complex tasks, how to effectively manage and optimize the behavior of these systems has become a core issue. Traditional software development has mature engineering practices—design patterns, code reviews, continuous integration—but agent system development is still in a relatively early stage.
Claude Code, as an AI coding assistant launched by Anthropic, provides strong customizability through mechanisms such as context files (CLAUDE.md), rules, skills, and hooks. However, the effective organization and maintenance of these components require a systematic methodology. The Reflexive Claude Code project was born in this context; it provides a complete framework and toolset for building, analyzing, and maintaining high-quality Claude Code agent systems.
Section 03
The core concept of Reflexive Claude Code is Agentic Context Engineering (ACE). This concept emphasizes that through carefully designed context structures and component interactions, AI assistants can more effectively understand tasks, perform operations, and continuously improve.
The project's name "Reflexive" reflects its key feature—the system can not only perform tasks but also self-analyze and optimize. This reflexivity is reflected in multiple levels:
Component-level self-verification: Each component (skill, rule, CLAUDE.md) is checked by a dedicated review agent to ensure quality meets standards.
System-level health check: The entire agent system is scanned through analysis skills to identify potential architectural flaws and optimization opportunities.
Process-level continuous improvement: The reflecting skill extracts lessons from the execution process and converts them into new rules or skills.
Section 04
Reflexive Claude Code defines a six-stage standardized workflow, with each stage handled by specialized skills:
Section 05
The migrating-agent-systems skill is responsible for assessing the project's maturity and routing it to the correct workflow. Maturity is divided into four levels:
For projects with existing rule files (e.g., .cursorrules), the migration skill also provides conversion suggestions to help users migrate configurations from other AI assistants to the Claude Code ecosystem.
Section 06
The analyzing-agent-systems skill performs a comprehensive health check on the existing system, covering 11 key categories:
| Category | Key Check Items |
|---|---|
| Routing/Triggers | Vague descriptions, overlapping triggers, missing handovers |
| Context Management | CLAUDE.md too large, greedy loading, context pollution |
| Workflow Continuity | Broken links, missing validation gates |
| Redundancy/Conflict | Duplicate rules, conflicting instructions |
| Security/Security | Unprotected sensitive files, excessive permissions |
| Observability | Lack of structured output, opaque routing |
| Architecture/Scalability | Flat topology, over-orchestration |
| Constitutional Stability | Procedural content in CLAUDE.md, vague instructions |
| Project Context | Missing deployment documentation, language coverage gaps |
| Cross-Tool Migration | Unimported .cursorrules, copilot instructions |
| Rule Health | Over 50 lines, missing paths:, dead globs, session startup over 300 lines |
The analysis results are presented in the form of a Rules Health Summary, providing a basis for subsequent optimization.
Section 07
The brainstorming-workflows skill explores the project's workflow through a complexity ladder (L1-L6) and maps it to patterns defined by Anthropic (such as Router, Orchestrator-Workers, Evaluator-Optimizer, etc.).
The complexity ladder includes:
Section 08
The planning-agent-systems skill adopts an architecture-first approach: first create a workflow diagram, then develop a component creation plan based on dependencies. This method ensures coordination and consistency between components.
Keep going with more reads from the same topic.
Splinter is a minimalist, high-performance key-value (KV) and vector storage system enabling zero-latency inter-process communication via shared memory and atomic operations. With only 766 lines of core code, it supports millions of operations per second and 768-dimensional vector storage, offering a new architectural approach for local LLM inference and data-intensive applications.
Folkering OS is the world's first AI-native bare-metal operating system, entirely written in Rust no_std without relying on Linux, POSIX, or libc. It can generate commands from scratch, compile them into WASM, and run them in 10 seconds, achieving true self-evolution.
This project explores how to use Large Language Models (LLMs) to detect logical vulnerabilities in Ethereum smart contracts, especially business logic flaws that are difficult for traditional static analysis tools to capture, such as economic manipulation and execution order errors.
A self-hosted MCP server that provides cross-session persistent coordination, task management, event bus, and observability for AI agents. It supports 35 tools and can be integrated into clients like Claude Code and Cursor with zero code changes.