AI agent guides.
For people who also use Claude Code, Codex, Gemini CLI or Copilot: agent memory, what goes in CLAUDE.md, MCP servers versus a CLI, and working in an Obsidian vault.
- CLI vs MCP for AI agents: context cost, auth and when to use each
- CLI vs MCP for AI agents: what tool definitions and results cost in Claude Code's context, how auth and sandboxes differ, and when MCP is the better pick.
- How to give AI agents persistent memory, and what each agent remembers
- Why agents forget between sessions, what Claude Code, Codex, Gemini CLI and Copilot keep natively, and how to give AI agents persistent memory across tools.
- How to use Claude Code with your Obsidian vault
- Use Claude Code with your Obsidian vault: run it in the vault folder, add a CLI for hash-checked edits and link-safe renames, and set permission rules.
- How to sync CLAUDE.md and Claude Code settings between machines
- How to sync CLAUDE.md between machines with skills, settings and plugins: what to move, what never to copy, Windows symlink traps and a setup for teams.
- Vector database vs markdown files for agent memory
- Vector database vs markdown for agent memory, plus SQLite FTS5, MCP memory servers, Mem0, Letta and Zep: what each stores, who can fix it, what it costs.
- What to put in CLAUDE.md, with a full example and MEMORY.md explained
- What to put in CLAUDE.md, where every file loads, an annotated example, a bloated file trimmed down, and how Claude Code's MEMORY.md auto memory works.