Client calls one alias
Agents use `coding-auto` through the OpenAI-compatible API.
Execution-aware routing for coding agents
An OpenAI-compatible routing brain that classifies each coding step, chooses the cheapest capable model, escalates when execution feedback says the work is risky, and uses local CodeGraph impact signals before spending premium tokens.
CodeRoute optimizes cost per successful software task by watching step type, budget, prior failures, repo impact, provider health, and feedback from tests or users.
Agents use `coding-auto` through the OpenAI-compatible API.
The request becomes a coding step with difficulty and risk.
Optional local analysis spots broad or high-impact changes.
Provider routing stays normalized across cloud and local models.
Failures and outcomes shape the next model decision.
The system keeps lightweight tasks on inexpensive or local capacity, while security reviews, architecture decisions, multi-file refactors, and high-impact repo changes route up faster.
Existing coding tools can point at `/v1`.
Local, cheap, balanced, and strong models.
CodeGraph runs locally against an index.
Small, inspectable pieces with clear jobs: API, ledger, provider gateway, deploy path, edge route, and repo intelligence.
Serves the OpenAI-compatible API, feedback endpoints, health checks, and this page.
Stores sessions, routing decisions, feedback, cost estimates, and model registry data.
Normalizes model calls across OpenRouter, direct APIs, and local model servers.
Provides optional local impact signals from a pre-indexed repository graph.
Pulls the GitHub repo and redeploys the Docker compose stack on push.
Publishes `coderoute.pchomelab.com` to the pchomelab host without opening inbound ports.
Human landing page on `/`, machine metadata on `/meta`, and OpenAI-compatible model access under `/v1`.