The Open-Source AI Stack
RSS

The Stack · Agents · Proprietary

Claude Code (Anthropic)

Anthropic's first-party CLI coding agent; closed-source; subscription model; the dominant closed coding agent in 2026.

Proprietary · stable · Project site →

Claude Code is Anthropic's first-party coding agent. CLI-based (with IDE integrations), proprietary, distributed primarily via the Anthropic Claude subscription tiers (Pro, Max, Team, Enterprise). MCP-native: integrates with the same MCP server ecosystem that the open agents use. Claude Code matters because it is the dominant closed coding agent in 2026 by deployment count among professional developers and the most direct demonstration of what a frontier-lab-built, tightly-integrated agent can do. Compared to siblings: Cursor (closed IDE with agent mode, large user base), Devin (Cognition Labs, fully autonomous pitch), Replit Agent / v0 / Lovable (closed product agents). Open counterparts: Goose, OpenHands, Aider, Cline, Continue. The structural difference is what the lock-in vector essay describes: Claude Code owns the workflow, the accumulated context (it remembers your codebase patterns), and the model-substitution decision (you cannot point it at a non-Anthropic model). Production-ready and widely deployed. Anthropic's "Claude for Open Source" program offers eligible OSS maintainers free Claude Max access. Strategic significance for the open-source-AI conversation: Claude Code is the strongest existence proof that a closed agent can win on user experience against an open agent at the same model-capability tier. The open-agent ecosystem competes on substitutability and sovereignty, not feature parity.

Sources

Want a follow-up? Ask the chat about Claude Code (Anthropic) in context. It will compare to siblings at the same layer and ground every claim in the wiki.

Other projects at the Agents layer

19 siblings · ordered open first