Dynamic Model Routing — AI skill for Claude Code
Risk-aware dynamic model routing for Codex CLI and Claude Code CLI.
How to install Dynamic Model Routing
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open iumi-in/dynamic-model-routing and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Dynamic Model Routing does
Risk-aware dynamic model routing for Codex CLI and Claude Code CLI.
Alternatives in AI
- Oh My Claudecode — 32 specialized agents and 7 execution modes for Claude Code, with smart model routing and automatic paralleliz 10.9k ★
- Launch An AI Feature — Workflow recipe — take an AI/LLM feature from a probabilistic-aware PRD to a launch-ready model card by chaini 1.3k ★
- Loushang — AI-native agent harness for coding workflows by python: multi-model LLM orchestration, stateful sessions, tool 1.3k ★
README
Dynamic Model Routing
[](https://github.com/iumi-in/dynamic-model-routing/actions/workflows/validate.yml) [](LICENSE)
Route coding tasks to the least expensive model that clears the risk bar, while preserving explicit model choices and verification.
Dynamic Model Routing is a native pilot plugin for **Codex CLI and Claude Code CLI**, based on the author's original skill. It classifies work as Clerical, Routine, Sensitive, or Escalation, then uses the host's eligible model-aware workers when available.
The plugin has no daemon, API proxy, API keys of its own, telemetry endpoint, or runtime dependency. It does not automatically switch the active coordinator, enforce monetary caps, or promise savings. Python is needed only for repository checks.
Support
| Host | Status | Current evidence |
|---|---|---|
| Codex CLI | Pilot | Installation, discovery, and removal verified |
| Claude Code CLI | Pilot | Strict manifest validation and session loading verified |
| Cursor and desktop-specific integrations | Roadmap | No support claim until a native adapter is tested |
Independent validation is the current priority. Contributors can verify [Codex routing on Linux](https://github.com/iumi-in/dynamic-model-routing/issues/1), verify [Claude Code loading on macOS](https://github.com/iumi-in/dynamic-model-routing/issues/2), or help collect the [first comparable routing benchmark](https://github.com/iumi-in/dynamic-model-routing/issues/3).
Install in 60 seconds
You need an installed, authenticated coding CLI with native skill and subagent support.
Related Skills
Cab
One local gateway for every coding agent. Cost-aware LLM model routing for Claude Code, Codex, OpenCode, and m
Lupin
Run Claude Code on any LLM provider without losing your setup, local-first proxy with verified profiles, OAuth
Arbiter
A local proxy that lets Claude Code use NVIDIA NIM models instead of Anthropic's API. Features task-aware mode
Paseo Agent Link
Account-aware routing, quota recovery, model visibility, and RAM protection for Paseo.
Agentic Engineering
Operate as an agentic engineer using eval-first execution, decomposition, and cost-aware model routing.
Token Station
Local routing control plane for AI agents and LLM providers, with a loopback-only gateway, smart and quota-awa
Related Agents
Codex CLI Research Agent
Senior OpenAI researcher specializing in Codex CLI - researches latest context window and model availability
Force Runner
Run AI model tasks in background via The Force. Use when user wants parallel work with GPT-5.2, Gemini, Codex,
Codex Runner
codex CLI(codex exec / codex exec resume)を最後まで走り切らせる専任エージェント。呼び出し元(メイン Claude / スキル)からプロンプト・model・effort・sandb