Anthropic says OpenClaw-style Claude CLI usage is allowed again
OpenClaw users can now directly integrate Claude CLI, simplifying agent deployment without API keys.
Anthropic has reversed its previous stance and now explicitly allows OpenClaw to reuse the Claude CLI (Command Line Interface) for powering AI agents. This policy shift, confirmed directly by Anthropic staff, means OpenClaw can treat `claude -p` usage and CLI session reuse as sanctioned for its integration. Developers running OpenClaw on hosts with an existing Claude CLI installation can now have their agents directly leverage that authenticated session, streamlining setup and reducing dependency on separate API key management for certain workflows.
For production environments, Anthropic still recommends using official API keys for the "clearest and most predictable" billing path. However, the CLI reuse option provides significant flexibility for development, testing, and internal tooling. The integration supports the latest Claude 4.6 models with features like adaptive thinking, which OpenClaw enables by default. It also includes support for Anthropic's prompt caching via the API to reduce costs and latency on repeated prompts, configurable with `cacheRetention` parameters set to 'short' (5 minutes) or 'long' (1 hour).
This move aligns with a trend of AI platforms offering more deployment flexibility. OpenClaw notes it also supports similar subscription-style CLI integrations with other providers like OpenAI Codex and Qwen Cloud, giving developers a unified interface for multiple AI backends. The change primarily benefits teams that have standardized on Claude's CLI for coding assistance and now want to incorporate those capabilities into automated agentic workflows managed by OpenClaw.
- Anthropic staff confirmed OpenClaw's method of reusing Claude CLI sessions is allowed, reversing a prior restriction.
- Integration supports Claude 4.6 models with adaptive thinking and API-based prompt caching (short/long retention).
- Provides a simplified auth path for development, though API keys are still recommended for clear production billing.
Why It Matters
Reduces friction for developers building Claude-powered agents, enabling faster prototyping and integration with existing CLI toolchains.