Topic archive
Software development lessons, implementation notes, and debugging writeups.
Explore the best practices for using Claude Code, an AI-powered coding assistant that transforms vibe coding into agentic engineering. Learn how to leverage commands, agents, skills, and hooks for efficient software development workflows.
Claude Code integrates OpenTelemetry metrics, enabling users to monitor coding activity and efficiency through tools like Grafana dashboards.
CodeGraphContext is a tool that indexes local code into a graph database, enabling AI assistants and developers to query and understand complex codebases through natural language and CLI commands.
Design.md is a format specification that enables coding agents to understand and apply a visual identity consistently through machine-readable design tokens combined with human-readable rationale.
One Skill to Rule Them All is a markdown-first meta-skill that observes work sessions, captures corrections and recurring patterns, and turns them into improvements for an agent skill library.
Claurst is a Rust terminal coding agent with strong early adoption, a GPL license, prebuilt releases, and a fast-moving implementation aimed at Claude-style coding workflows.
Tmux Orchestrator is a Python and shell project for coordinating Claude agents inside tmux sessions so multiple agent roles can keep working across projects and time.
Agent Rules Books is a markdown library of compact, ready-to-use coding-agent rules distilled from classic software engineering books for Codex, Claude Code, Cursor, and similar tools.
LLM Wiki Compiler is a TypeScript CLI that turns raw sources into an interlinked markdown wiki, giving the LLM Wiki pattern a more productized compiler-style workflow.
Obsidian Wiki is a young but fast-moving Python and markdown framework for letting AI agents build and maintain an Obsidian knowledge base around the LLM Wiki pattern.
Everything Claude Code is a large agent-harness optimization system for Claude Code and adjacent tools, covering skills, instincts, memory, security, MCP, and research-first workflows.
OpenCode Chrome Annotation is a young TypeScript project that pairs an OpenCode plugin with a Chrome extension so you can select a live page element, attach an instruction, and send browser context plus a screenshot into an active OpenCode session.
Matomo is a mature GPL web analytics platform, formerly Piwik, that offers a self-hosted and privacy-focused alternative to Google Analytics with a long-running PHP/MySQL codebase.
RabbitLLM is a young Python package that adapts AirLLM-style layer streaming so very large Qwen models can run on small consumer GPUs by loading one layer at a time.
Rybbit is a fast-growing, open-source web and product analytics platform that tries to replace Google Analytics with a privacy-friendly, self-hostable, and more approachable experience.
daily_stock_analysis is a popular Python-based AI stock analysis system for A-share, Hong Kong, and U.S. markets, combining market data providers, news search, LLM-generated decision dashboards, web UI, bots, Docker, and scheduled GitHub Actions runs.
ComposioHQ's awesome-codex-skills is a fast-growing curated library of Codex skill folders, spanning code workflows, productivity automation, communication, data analysis, and meta utilities.
Matt Pocock's skills repository is a highly popular, MIT-licensed library of Claude-style agent skills that encodes practical engineering workflows such as diagnosis, TDD, issue triage, architecture review, and planning interviews.
Agent of Empires is an actively maintained Rust session manager that wraps tmux, git worktrees, optional Docker sandboxing, and a browser dashboard around multi-agent coding workflows.
A source-backed look at jcode, a Rust coding-agent harness built around fast terminal UX, multi-session workflows, and extensible agent tooling.
A book-note style reflection on extracting durable practice from technical reading.
Using public security guidance as a source-backed way to structure application security learning.
A practical note on how AI can help shape a draft without replacing accountability for the final post.
A launch note on the site architecture, editorial constraints, and why the stack is intentionally boring.