Claude Code

Summary

Claude Code is Anthropic's official CLI for Claude — the AI assistant that drives ORE Studio's day-to-day development. It runs inside the repository, reads CLAUDE.md at session start, and has access to a curated set of tools and permissions defined in Claude Code Settings. Project-specific configuration lives under .claude/; skills, runbooks, and recipes give Claude Code the context it needs to act without hand-holding at each step.

Detail

What Claude Code is

Claude Code (claude CLI) is Anthropic's interactive agent for software engineering tasks. Unlike a plain chat interface, Claude Code operates directly on the filesystem — it reads files, edits code, runs shell commands, and opens pull requests. It is invoked from the repository root; CLAUDE.md is the mandatory entry point it reads at the start of every session.

How it is used in ORE Studio

ORE Studio uses Claude Code as its primary development tool. Humans set goals, review work, and make architectural decisions; Claude Code executes — writing code, authoring documentation, opening PRs, and responding to review comments. The agile process (sprint → story → task → PR) is the handoff boundary: a human approves a story's acceptance criteria, and Claude Code implements and delivers it.

Session configuration is layered: .claude/settings.json controls the tool permission allow-list and sandbox overrides; .claude/skills/ contains task-shaped playbooks; doc/llm/runbooks/ contains multi-step procedures; and doc/recipes/ holds individual how-tos with executable shell blocks. Claude Code reads these as needed to orient itself and act correctly.

Configuration

The permission allow-list and sandbox overrides are maintained as a literate org document and generated via org-babel tangle. Never edit .claude/settings.json directly; regenerate it with the deploy_settings CMake target. Full details in Claude Code Settings.