Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Claude Code Model Router: Delegate Work to Cheap Models

    2

    Save your Claude limits by dispatching token hungry tasks to cheaper subagents like GLM, Kimi, DeepSeek etc.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Claude Code Model Router: Delegate Work to Cheap Models

    Claude Code Model Router: Delegate Work to Cheap Models

    Example session with this skill installed

    I want to set up a worker for DeepSeek via OpenRouter so I can offload some boilerplate coding tasks without hitting my Anthropic limits.

    • Read your context and instructions
    • Compiled the claude code model

    I will help you set up a DeepSeek worker. We'll use a dedicated configuration directory to ensure your OpenRouter credentials don't interfere with your primary Anthropic session.

    DeepSeek Worker Setup

    To initialize the worker, run the following commands in your terminal:

    
    # 1. Create the worker using the OpenRouter preset
    "${CLAUDE_PLUGIN_ROOT}/scripts/setup-worker.sh" create deepseek --provider openrouter
    
    # 2. Add your OpenRouter API key (this avoids leaking it in your shell history)
    # Replace 'sk-or-...' with your actual key
    printf 'sk-or-your-api-key' | "${CLAUDE_PLUGIN_ROOT}/scripts/setup-worker.sh" set-key deepseek
    
    # 3. Finalize the headless configuration
    "${CLAUDE_PLUGIN_ROOT}/scripts/setup-worker.sh" finalize deepseek
    

    Once finalized, you can delegate tasks to this worker. The worker will operate in its own sandbox at ~/.claude-workers/deepseek/.

    Example dispatch command:

    ~/.claude-workers/deepseek/run -p "Write standard CRUD boilerplate for the user_profile model in lib/models.go" --permission-mode acceptEdits
    

    Next steps

    • Verify the connection by running ~/.claude-workers/deepseek/run to start an interactive test session.
    • Review ~/.claude-workers/deepseek/settings.json if you need to manually adjust model mappings (e.g., pointing Sonnet to deepseek/deepseek-chat).
    • Add the delegation rules from templates/delegation-rules.md to your CLAUDE.md to improve worker reliability.

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

    About this skill

    Manager/worker orchestration for Claude Code: keep your smartest model as the manager and delegate implementation to cheap subagents and workers on z.ai GLM, Kimi, DeepSeek, Ollama, or any Anthropic-compatible API — and see exactly how many tokens you save.

    Requires Anthropic-compatible endpoints. MLX-variant models may hang on local Ollama setups. Environment variables in the parent shell can override worker settings if not using the run wrapper.

    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 any payment.

    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

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    Permissions

    Terminal / Shell
    Network Access
    Listed1 month ago

    What's inside

    Frequently Asked Questions

    This skill functions as a management layer for Claude Code, allowing you to route high-volume tasks like boilerplate generation or unit testing to lower-cost models while reserving your primary Claude limits for complex reasoning.

    The plugin supports any Anthropic-compatible API, meaning you can easily connect it to providers like DeepSeek, LiteLLM, Ollama, Kimi, and z.ai GLM.

    Setup is straightforward: once installed via the marketplace, you simply provide your worker API keys and endpoints within your Claude Code configuration to start delegating tasks.

    You will receive the core orchestration plugin files, a configuration guide for various API providers, and access to all future version updates to ensure ongoing compatibility.

    Yes, you can define specific rules to determine which types of tasks are sent to headless workers and which remain with the primary Claude model to ensure optimal performance.

    New in Agents & Orchestration