Claude Code#

An agentic coding tool by anthropic that lives in your terminal, understands your codebase, and helps you code faster through natural language commands.

Architecture#

Claude Code is the “agentic harness” around Claude models. It provides tools, context management, and execution environment that turn a language model into a coding agent.

The Agentic Loop#

Three phases: gather context → take action → verify results, chained together with course-correction. Claude decides what each step requires based on what it learned from the previous step.

Five Tool Categories#

File operations — read, edit, create, rename
Search — find files by pattern, regex content search
Execution — shell commands, servers, tests, git
Web — search, fetch docs, look up errors
Code intelligence — type errors, definitions, references (via plugins)

Three Execution Environments#

Local: Your machine (default)
Cloud: Anthropic-managed VMs
Remote Control: Your machine, controlled from browser

Memory#

Dual system for cross-session knowledge:

CLAUDE.md: Human-written instructions. Scoped to org/project/user/local. Loaded at session start. Target under 200 lines. Supports @imports and .claude/rules/ for path-specific rules.
Auto memory: Claude-written learnings (build commands, debugging insights, preferences). Per working tree. First 200 lines or 25KB loaded at session start.

Reads CLAUDE.md, not AGENTS.md — but can import AGENTS.md for cross-tool compatibility.

Permission Modes#

Mode	Auto-approved	Best for
`default`	Reads only	Sensitive work
`acceptEdits`	Reads + file edits	Iterating on code
`plan`	Reads only (research mode)	Exploring before changing
`auto`	Everything (classifier-reviewed)	Long tasks
`dontAsk`	Pre-approved tools only	CI/CD
`bypassPermissions`	Everything	Containers/VMs only

Extensibility#

Five extension mechanisms:

mcp-protocol — Open standard for external tool integration
Plugins — Custom commands and functionality
Skills — SKILL.md files following agent-skills-standard. Bundled: /batch (parallel changes in git worktrees), /simplify, /loop, /debug. Plugin marketplace for distribution.
Hooks — 25+ lifecycle events. Shell commands, HTTP endpoints, or LLM prompts. Can block actions, inject context, automate workflows.
Subagents — Own context window, system prompt, tool access. Built-in: Explore (Haiku), Plan, General-purpose. Custom via markdown files. Cannot spawn sub-subagents.

Platforms#

Terminal (CLI), VS Code, JetBrains, Desktop, Web, Chrome extension, Slack, GitHub (@claude on PRs), GitHub Actions, GitLab CI/CD, Remote Control, iOS app.

Sessions portable across all surfaces. Same CLAUDE.md and MCP servers everywhere.

In the Ecosystem#

Built by anthropic, powered by Claude models
One of the harnesses supported by scion (as claude harness)
Competes with / complements kiro (AWS), Gemini CLI (Google), Codex (OpenAI)
AGENTS.md import enables cross-tool instruction sharing