Technology · Agentic → Coding Agents · Commercial

Claude Code

Taxonomy

Claude Code is Anthropic’s agentic coding tool: a session over a project that reads/edits files, runs commands, and uses tools under permission modes — in the terminal, IDE, desktop app, and browser. It is not a multi-channel personal assistant. The brain is Claude API (Claude models on your plan or API key). Peers in this cell: Continue (open-core IDE), Aider (OSS CLI), OpenHands, SWE-agent.

DocsStack: agentic-coding-agents

Why it matters here

Coding agents are high-signal because git/CI already falsify claims. Claude Code is the Anthropic-shaped instance of tech-coding-agents: repo loop + tools + subscription/API spend. Pair with OpenClaw only when you need standing chat channels and hosted gateway policy — OpenClaw does not replace Claude Code. Keep side effects behind human or policy gates.

How it works

You run claude in a repo. The agent assembles context, proposes tool calls, waits on permission/hooks where configured, applies edits, and continues until the task stops. Model choice and effort follow Claude API rules (Opus 5.5 = adaptive thinking, no forced tools).

Related: Claude API · Continue · Aider · OpenClaw · tech-coding-agents · tech-code-sandbox

Flow

When to reach for it

  • Use when: the job is terminal/repo-shaped coding with Claude as the model and you want Anthropic’s CLI agent loop.
  • Skip when: you need IDE-native chrome — prefer Continue; or issue-autonomy harnesses — prefer OpenHands / SWE-agent; or WhatsApp/Telegram front door — prefer OpenClaw in addition, not instead.
  • Prefer instead: Continue for IDE; Aider for open-source CLI; Claude API alone when you own the agent loop.

Limits

  • Vendor lock on the brain: quality tracks Claude releases; Opus 5.5 migration breaks (thinking, tools, computer toolset) hit any wrapper that still forces tools or disables thinking.
  • Permission surface: unrestricted shell is remote code execution on the developer machine — treat like tech-code-sandbox risk.
  • Scope: coding agent ≠ company OS, ledger, or approval authority.

What we checked

Claims below are backed by science sources on disk.

Product overview

Claude Code docs · STRONG

“Claude Code is an agentic coding tool that reads your codebase, edits files, runs commands, and integrates with your development tools. Available in your terminal, IDE, desktop app, and browser.”

GitHub product CLI

Claude Code GitHub · MODERATE

“Claude Code is an agentic coding tool that lives in your terminal”

Autonomy / agent framing

Claude Code autonomy note · MODERATE

“Claude Code can now work more independently on complex coding tasks”
Research inventory

9 tags · 5 out · 6 in · 3 artifacts · 1 gaps · 0 corpus docs

Catalog tags

landscape.layer
Agentic
landscape.subcategory
Coding Agents
license_tag
Commercial
maps.dm
absent
maps.features
0
one_liner
Agentic
review.depth
science
slug
claude-code
title
Claude Code

Artifacts

  • dm_map · absent
  • features_map · present · technologies/claude-code/features.md
  • readme · present · technologies/claude-code/README.md

Out · alternative_to

In · alternative_to

In · in_stack