Model Router for Codex

    1

    Choose economical Codex models and reasoning levels, request supported model changes, and escalate when task evidence calls for stronger reasoning.

    Free

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Model Router for Codex

    Example session with this skill installed

    Rewrite this sentence so it is clearer: “The meeting was moved due to a clash in schedules.”

    • Read your context and instructions
    • Compiled the model router

    Illustrative example

    Progress update
    Requested GPT-6 Luna with low reasoning for the continuation because this is a short editing task.

    Final response
    The meeting was rescheduled because of a scheduling conflict.

    The progress notice reports an accepted routing request. A confirmed switch notice is used when runtime metadata identifies the selected model. The final response completes the original editing request.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Model Router helps Codex choose a suitable model and reasoning level for each task. It provides routes for writing, coding, research, browser operations, documents, data and media workflows.

    It requests model changes through controls exposed by the host, including same-chat continuations where supported. Routing notices distinguish accepted requests from model changes confirmed by runtime metadata. The original task remains the deliverable after a handoff.

    The final 10% of remaining allowance triggers Luna economy mode when a reliable usage signal is available, unless the user explicitly overrides it. Model availability, usage visibility and switching controls depend on the host.

    The package includes SKILL.md, Codex agent metadata and a README with installation instructions, category routes, an illustrative cost comparison and a demo exchange. Source and updates are available at https://github.com/lockej2005/model-router-skill.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.

    Cannot intercept the currently executing turn. Switching depends on runtime support for thread continuations. Accuracy of allowance tracking relies on host-provided usage metadata.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 7 days ago

    • Free with an Agensi account

    Software
    Codex with access to supported modelsGit

    Permissions

    Read Files
    Listed7 days ago

    What's inside

    Frequently Asked Questions

    It chooses economical model and reasoning routes for the task and requests supported model changes. Savings depend on context, tool use, retries and the host’s billing rules.

    It is designed for Codex with available GPT-6 family models. Automatic changes require a supported host model-selection control, such as a same-chat continuation override.

    When the host exposes a reliable remaining-allowance signal of 10% or less, the instructions select Luna at low reasoning unless the user explicitly overrides it. Usage visibility depends on the host.

    The package includes SKILL.md, Codex agent metadata and a README with installation guidance, category defaults and an illustrative demo.

    The instructions call for checking results, making one focused correction where appropriate, and escalating based on observed failures. The agent follows the tools and permissions available in its host.

    New in Agents & Orchestration

    Free