goose-headroom-provider

Configure Goose Custom Provider for Headroom

Create and configure a Goose custom provider that points to the local Headroom context-optimization proxy. This skill covers only the Goose provider configuration — for Headroom installation and systemd setup, see the headroom-litellm-proxy skill.

Traffic Chain

Goose (custom_headroom) → Headroom Proxy (:8787) → LiteLLM (:4000) → Vertex AI (Claude)
                           ↑ context compression        ↑ model routing

Prerequisites

• Goose installed (goose CLI or Goose Desktop)
• Headroom proxy running on port 8787 (use headroom-litellm-proxy skill to set up; verify with headroom-proxy-status skill)
• LiteLLM proxy running on port 4000 (upstream of Headroom)

---

Reference Configuration

Custom Provider JSON

File: ~/.config/goose/custom_providers/custom_headroom.json

{
  "name": "custom_headroom",
  "engine": "openai",
  "display_name": "Headroom",
  "description": "Headroom context-optimization proxy → LiteLLM (Vertex AI Claude models)",
  "api_key_env": "",
  "base_url": "http://host.containers.internal:8787",
  "models": [
    {
      "name": "claude-opus-4-6",
      "context_limit": 128000,
      "input_token_cost": null,
      "output_token_cost": null,
      "currency": null,
      "supports_cache_control": null,
      "reasoning": false
    },
    {
      "name": "claude-sonnet-4-6",
      "context_limit": 128000,
      "input_token_cost": null,
      "output_token_cost": null,
      "currency": null,
      "supports_cache_control": null,
      "reasoning": false
    },
    {
      "name": "claude-sonnet-4-5",
      "context_limit": 128000,
      "input_token_cost": null,
      "output_token_cost": null,
      "currency": null,
      "supports_cache_control": null,
      "reasoning": false
    }
  ],
  "headers": null,
  "timeout_seconds": 600,
  "supports_streaming": true,
  "requires_auth": false,
  "catalog_provider_id": null,
  "base_path": null,
  "env_vars": null,
  "dynamic_models": null,
  "skip_canonical_filtering": false,
  "model_doc_link": null,
  "setup_steps": [],
  "fast_model": null,
  "preserves_thinking": true
}