docs: clarify Claw Code positioning as multi-provider Claude-Code-shaped runtime

[Yeachan-Heo]

Source

Discord #claw-code product questions: users asked whether Claw Code is intended only for Claude/Anthropic models while OpenCode is for Codex/Kimi/other providers.

Problem

The name and Claude-Code-shaped UX make people assume Claw Code is Claude-only. In practice, the product direction appears to be a Claude-Code-like workflow/runtime that can target non-Anthropic/OpenAI-compatible/local providers, with some rough edges.

Suggested docs/README wording

Clarify:

• Claw Code is not intended to be Claude-only.
• It is a Claude-Code-shaped workflow/runtime.
• It can target Anthropic and OpenAI-compatible/provider-routed/local models depending on config.
• Non-Claude providers may require stricter response-shape/tool-call compatibility.
• Provider-specific identity leaks are bugs, not product intent.

Acceptance criteria

• README or docs answer “Is this Claude-only?” directly.
• Provider support is framed honestly: supported, but contracts differ.
• Users understand when OpenCode may be the more polished daily-driver today vs when Claw Code is useful for runtime/provider hackability.

—
[repo owner's gaebal-gajae (clawdbot) 🦞]