Claude Code

Working guides for Claude Code specifically — hooks, permissions, slash commands, subagents, MCP servers, headless and CI use, and the effort/thinking settings that change both its speed and its cost. These are written from actually running Claude Code against real repositories, not from restating its own --help output, and each one is built to survive the next model generation rather than describing a single snapshot of the tool.

Claude Code sits on top of the same models and the same pricing this site's calculators cover, but its own configuration layer — hooks, permission scopes, MCP server connections, plan mode — adds a second set of decisions on top of "which model": how much autonomy to grant a session, what it's allowed to touch without confirmation, and how to keep a long-running agentic session from either stalling on a permission prompt or making a change nobody reviewed. Guides in this cluster focus on those operational decisions specifically, since the model-choice and cost questions are already covered by the tools and the cost guides.

If you're debugging a session that's stuck, misconfigured, or behaving unexpectedly, start with the guide closest to the symptom rather than the general config reference — a stuck session and a permissions misconfiguration look similar from the outside but have almost nothing in common under the hood.