Unify your AI coding tool workflow
Provider switching · local routing · token stats · auto failover for mainstream AI CLIs
ChatGPT, Codex, Claude Code, OpenClaw, Harness, and Grok CLI each need their own API Key
Unclear how much each model costs, no way to optimize
429 errors or quota exhaustion interrupt your workflow
Hard to find the best value among different providers
Track usage and cost by provider and model
Local routing + virtual model groups with priority failover
ChatGPT, Codex, Claude Code, OpenClaw, Harness, Grok CLI and other AI CLIs
API Keys stored in local SQLite, no third party
Point every AI coding tool to the local routing gateway for provider switching, format conversion, failover, and usage tracking
AI tools point to the local gateway
Keys, formats, and logs stay local
Select model channels by priority and availability
Tokens, requests, and costs are recorded automatically
UI Preview
Virtual model groups aggregate multi-provider channels with priority-based auto failover; the Apps page takes over mainstream AI tools in one click
One Base URL · OpenAI Compatible · Auto-match providers · Real-time routing logs
Keywords: local gateway / OpenAI compatible / routing logs
Multi-channel aggregation · Drag to set priority · Real-time status monitoring
Keywords: priority / auto failover / model groups
ChatGPT · Codex · Claude Code · OpenClaw · Harness — take over in one click
Keywords: one-click takeover / provider switching / local routing
Unified multi-provider management · One-click add from preset library · Seamless OpenAI-compatible integration
Keywords: provider presets / API Key / custom Base URL
By model / provider / date / request · Make every token visible
Keywords: usage stats / cost tracking / request details
One-click add from preset library, just enter your API Key. Also supports any OpenAI-compatible provider
Open source under MIT license. Free for commercial use, modification, and distribution. All code is on GitHub.