Kimi Code

Connect Kimi Code to Routero so its model calls run through your gateway — attributed, budgeted, and logged — while Kimi Code keeps handling repository access and file edits locally.

Kimi Code connects through a custom provider speaking the OpenAI protocol.

What you must set

In Kimi Code, open Settings → Providers → Add provider → Custom:

Setting Value
Name Routero
API protocol OpenAI
API Key a Routero virtual key
Base URL https://api.routero.dcsmartvision.com/v1
Model ID a model Routero serves, e.g. deepseek/deepseek-v4-flash
Max context size the model’s context window, e.g. 128000
Display name anything recognizable, e.g. Routero DeepSeek V4 Flash

Save the provider, then select its model in Kimi Code and send a test prompt.

Add one provider entry per model you plan to use — each custom provider carries a single Model ID. Any model string Routero serves works.

Create a key for each developer

In the dashboard, open API Keys and create a virtual key per developer — scoped to their team, limited to the approved models, with an optional budget. Each developer’s Kimi Code traffic is then attributed individually, and requests appear in Logs routed through Routero rather than straight to the underlying provider.

Gateway policies on Kimi Code traffic

Policy-bound AI capabilities attached to the key apply to Kimi Code traffic:

  • Guardrails — enforced per request, including changes made mid-conversation: switching a guardrail from Monitor to Block takes effect on the next request, and Kimi Code surfaces the block clearly as 400 Request blocked by content filter.
  • Token saving — repeated prompts hit the exact cache; semantically equivalent rewrites hit the semantic cache.
  • Memory — a memory session bound through the key’s policy is read and written: stored facts are retrieved in new conversations, and new facts stated in a conversation are stored to Routero automatically.

→ Calling the API for the base URL and auth model. → Claude Code, ZCode, and Codex for the other agents.