Claude Code

Claude Code. Is Anthropic’s agentic coding assistant (terminal CLI, VS Code extension, and desktop surfaces) that reads your repo, edits files, runs commands, and follows project rules.

Summary

Claude Code runs an agent loop in your project context: gather context (files, search, MCP), act (edit, bash, git), verify (tests, diffs). It supports hooks, subagents, plan mode, and the Agent Skills Framework directory layout used across this vault (SKILL.md files).

Use it for interactive development, refactors, and skill-driven workflows when Claude models or Anthropic billing are the constraint. It is a repo-bounded AI Agent; not an omnichannel personal bot. Always-on messaging gateways (OpenClaw, Hermes) are hold in this subcategory. For scheduled automation, prefer deterministic scripts and Dagu with LLM steps only where reasoning is required (projects/research/morning-briefing-personal-automation-platform).

Compare Codex for the same agent-loop-in-repo pattern on OpenAI plans.

Details

• Surfaces: terminal (claude in repo root), VS Code extension, desktop app; pick one primary surface per machine.
• Skills: load from .claude/skills/ or shared agent-skills/ repos; see Agent Skills - Sources for curated lists.
• Auth: Claude subscription or Anthropic Console API; some surfaces allow third-party model routing.
• Fit: Tool / AI Agent, agent-loop coding tool in the workspace (IDE surfaces are deployment options, not the taxonomy).
• Contrast: OpenClaw / Hermes for omnichannel bots; Cursor + cursor-agent for our default editor workflow; Codex on OpenAI plans.