Some notes on AI Agent Rule / Instruction / Context files / etc.
A blazing-fast markdown document management system built with PostgreSQL and Fish shell. Handles everything from personal notes to millions of documents with smart tab completion and beautiful mdcat rendering.
- 🚀 Scales to millions - Optimized indexes, parallel queries, batch operations
- 🔍 Full-text + Fuzzy search - PostgreSQL FTS with trigram support
- 📁 Project organization - Group documents by project with multi-level tab completion
- ⚡ Sub-second queries - Even with millions of documents
ABSOLUTE RULE: ALL operations MUST be concurrent/parallel in a single message:
- TodoWrite: ALWAYS batch ALL todos in ONE call (5-10+ todos minimum)
- Task tool: ALWAYS spawn ALL agents in ONE message with full instructions
- File operations: ALWAYS batch ALL reads/writes/edits in ONE message
TL;DR
This guide provides a comprehensive overview of advanced strategies to maximize efficiency and control when using Claude Code, Anthropic's AI coding assistant. It covers foundational elements like setting up global rules (.claw.md
) for system-wide prompts and managing permissions via settings.local.json
to enable autonomous operation while maintaining security. The video then delves into agentic workflows through custom slash commands, enhancing capabilities with external tools via MCP servers (highlighting Serena for semantic code understanding and teasing Archon for knowledge/task management). Key strategies include context engineering with the PRP framework for structured project definition and execution, leveraging specialized sub-agents for delegated tasks, and implementing Claude hooks for automated actions. Finally, it explores integrating with GitHub CLI for automated issue fixing and PR creation, utilizing YOLO mode w
<system_context> You are an expert CLAUDE.md editor. IMPORTANT: If you make any changes that makes any CLAUDE.md file out of date, please update the CLAUDE.md file accordingly. </system_context>
<critical_notes>
-
Code with elegance - Write clean, maintainable, and elegant code that follows established patterns.
-
Follow the paved path - ULTRA CRITICAL: The
paved path
is the PREFERRED way of doing things. When you encounterpaved path
in any documentation, this indicates the canonical approach that MUST be followed.
A simple example showing how to create custom slash commands for Claude Code without needing MCP servers.
Claude Code automatically recognizes any markdown file in your .claude/commands/
directory as a slash command. This provides a simple way to create reusable prompts and workflows without any complex setup.
On Mac or Linux:
This document explains the mechanics of how CLAUDE.md files function in Claude Code, based on empirical testing and research.
- All CLAUDE.md files are refreshed between sessions - both root and directory-specific files pick up changes when resuming
- NOT refreshed during conversation - changes made during an active session aren't visible until a new conversation starts
- Resuming a session loads updated content - exiting and resuming Claude Code will pick up all CLAUDE.md changes
This script installs a comprehensive collection of Claude Code agents and commands to supercharge your development workflow.
This installer provides:
- Git Commit Action →
/git-commit
- Professional commits with Conventional Commits + emoji