feat: add OrcaRouter as a named LLM provider

README.md

@@ -77,6 +77,8 @@ Each agent has a `soul.md` (personality), `memory.md` (long-term memory), and a
 - Network access to LLM API endpoints
 
 > **Note:** Clawith does not run any AI models locally — all LLM inference is handled by external API providers (OpenAI, Anthropic, etc.). The local deployment is a standard web application with Docker orchestration.
+>
+> **OrcaRouter** is available as a first-class provider: in the enterprise LLM settings, add a model with provider **OrcaRouter** (base URL `https://api.orcarouter.ai/v1`) and any `sk-orca-...` key. See [orcarouter.ai](https://www.orcarouter.ai). It also runs gateway-level, zero-trust security for AI agents on the same endpoint — screening every prompt/response and governing every tool call on a default-deny basis, with no application code changes.
 
 #### Recommended Configurations
 

backend/app/services/llm/client.py

@@ -2376,6 +2376,13 @@ class ProviderSpec:
         default_base_url="https://openrouter.ai/api/v1",
         default_max_tokens=4096,
     ),
+    "orcarouter": ProviderSpec(
+        provider="orcarouter",
+        display_name="OrcaRouter",
+        protocol="openai_compatible",
+        default_base_url="https://api.orcarouter.ai/v1",
+        default_max_tokens=16384,
+    ),
     "zhipu": ProviderSpec(
         provider="zhipu",
         display_name="Zhipu",

backend/tests/test_orcarouter_provider.py

@@ -0,0 +1,55 @@
+"""Provider registry tests for the OrcaRouter integration."""
+
+from app.services.llm.client import (
+    PROVIDER_REGISTRY,
+    create_llm_client,
+    get_provider_base_url,
+    get_provider_manifest,
+    normalize_provider,
+)
+
+
+def test_orcarouter_registered_in_provider_registry() -> None:
+    spec = PROVIDER_REGISTRY.get("orcarouter")
+    assert spec is not None
+    assert spec.provider == "orcarouter"
+    assert spec.display_name == "OrcaRouter"
+    assert spec.protocol == "openai_compatible"
+    assert spec.default_base_url == "https://api.orcarouter.ai/v1"
+    assert spec.supports_tool_choice is True
+    assert spec.default_max_tokens > 0
+
+
+def test_orcarouter_manifest_round_trip() -> None:
+    manifest = {entry["provider"]: entry for entry in get_provider_manifest()}
+    entry = manifest.get("orcarouter")
+    assert entry is not None
+    assert entry["display_name"] == "OrcaRouter"
+    assert entry["default_base_url"] == "https://api.orcarouter.ai/v1"
+
+
+def test_orcarouter_base_url_resolution() -> None:
+    assert get_provider_base_url("orcarouter") == "https://api.orcarouter.ai/v1"
+    # explicit custom base_url takes precedence
+    assert get_provider_base_url("orcarouter", "https://example.com/v1") == "https://example.com/v1"
+    # aliases normalize to the canonical provider id
+    assert normalize_provider("OrcaRouter") == "orcarouter"
+
+
+def test_create_orcarouter_client_uses_openai_compatible_protocol() -> None:
+    client = create_llm_client(
+        provider="orcarouter",
+        api_key="sk-orca-test",
+        model="orcarouter/auto",
+    )
+    try:
+        assert client.base_url == "https://api.orcarouter.ai/v1"
+        assert client.model == "orcarouter/auto"
+        assert client.supports_tool_choice is True
+        # Bearer auth headers for the OpenAI-compatible gateway
+        headers = client._get_headers()
+        assert headers["Authorization"] == "Bearer sk-orca-test"
+    finally:
+        import asyncio
+
+        asyncio.run(client.close())

frontend/src/pages/enterprise-settings/tabs/LlmTab.tsx

@@ -57,6 +57,7 @@ const FALLBACK_LLM_PROVIDERS: LLMProviderSpec[] = [
     { provider: 'baidu', display_name: 'Baidu (Qianfan)', protocol: 'openai_compatible', default_base_url: 'https://qianfan.baidubce.com/v2', supports_tool_choice: false, default_max_tokens: 4096 },
     { provider: 'gemini', display_name: 'Gemini', protocol: 'gemini', default_base_url: 'https://generativelanguage.googleapis.com/v1beta', supports_tool_choice: true, default_max_tokens: 8192 },
     { provider: 'openrouter', display_name: 'OpenRouter', protocol: 'openai_compatible', default_base_url: 'https://openrouter.ai/api/v1', supports_tool_choice: true, default_max_tokens: 4096 },
+    { provider: 'orcarouter', display_name: 'OrcaRouter', protocol: 'openai_compatible', default_base_url: 'https://api.orcarouter.ai/v1', supports_tool_choice: true, default_max_tokens: 16384 },
     { provider: 'kimi', display_name: 'Kimi (Moonshot)', protocol: 'openai_compatible', default_base_url: 'https://api.moonshot.cn/v1', supports_tool_choice: true, default_max_tokens: 8192 },
     { provider: 'vllm', display_name: 'vLLM', protocol: 'openai_compatible', default_base_url: 'http://localhost:8000/v1', supports_tool_choice: true, default_max_tokens: 4096 },
     { provider: 'ollama', display_name: 'Ollama', protocol: 'openai_compatible', default_base_url: 'http://localhost:11434/v1', supports_tool_choice: true, default_max_tokens: 4096 },