feat: add Atlas Cloud LLM provider

backend/app/services/llm/client.py

@@ -2035,6 +2035,8 @@ class ProviderSpec:
 
 # Provider aliases accepted for compatibility
 PROVIDER_ALIASES: dict[str, str] = {
+    "atlas": "atlascloud",
+    "atlas-cloud": "atlascloud",
     "openai_response": "openai-response",
     "openairesponses": "openai-response",
 }
@@ -2064,6 +2066,13 @@ class ProviderSpec:
         default_base_url="https://api.openai.com/v1",
         default_max_tokens=16384,
     ),
+    "atlascloud": ProviderSpec(
+        provider="atlascloud",
+        display_name="Atlas Cloud",
+        protocol="openai_compatible",
+        default_base_url="https://api.atlascloud.ai/v1",
+        default_max_tokens=8192,
+    ),
     "azure": ProviderSpec(
         provider="azure",
         display_name="Azure OpenAI",

backend/tests/test_atlas_cloud_llm_provider.py

@@ -0,0 +1,88 @@
+"""Atlas Cloud provider registry coverage."""
+
+from __future__ import annotations
+
+from app.services.llm.client import (
+    LLMMessage,
+    OpenAICompatibleClient,
+    create_llm_client,
+    get_max_tokens,
+    get_provider_base_url,
+    get_provider_manifest,
+    get_provider_spec,
+    normalize_provider,
+)
+
+
+def test_atlas_cloud_aliases_resolve_to_canonical_provider() -> None:
+    assert normalize_provider("atlas") == "atlascloud"
+    assert normalize_provider("atlas-cloud") == "atlascloud"
+    assert normalize_provider("ATLASCloud") == "atlascloud"
+
+
+def test_atlas_cloud_provider_spec_and_manifest_defaults() -> None:
+    spec = get_provider_spec("atlas-cloud")
+
+    assert spec is not None
+    assert spec.provider == "atlascloud"
+    assert spec.display_name == "Atlas Cloud"
+    assert spec.protocol == "openai_compatible"
+    assert spec.default_base_url == "https://api.atlascloud.ai/v1"
+    assert spec.supports_tool_choice is True
+    assert spec.default_max_tokens == 8192
+    assert get_provider_base_url("atlas", None) == "https://api.atlascloud.ai/v1"
+    assert get_provider_base_url("atlascloud", "https://proxy.example/v1") == "https://proxy.example/v1"
+
+    manifest_entry = next(item for item in get_provider_manifest() if item["provider"] == "atlascloud")
+    assert manifest_entry["display_name"] == "Atlas Cloud"
+    assert manifest_entry["default_base_url"] == "https://api.atlascloud.ai/v1"
+    assert manifest_entry["supports_tool_choice"] is True
+    assert set(manifest_entry["aliases"]) == {"atlas", "atlas-cloud"}
+
+
+def test_atlas_cloud_factory_uses_openai_compatible_client() -> None:
+    client = create_llm_client(
+        provider="atlas",
+        api_key="test-key",
+        model="qwen/qwen3.5-flash",
+    )
+
+    assert isinstance(client, OpenAICompatibleClient)
+    assert client.base_url == "https://api.atlascloud.ai/v1"
+    assert client.model == "qwen/qwen3.5-flash"
+    assert client.supports_tool_choice is True
+    assert client._normalize_base_url() == "https://api.atlascloud.ai/v1"
+
+
+def test_atlas_cloud_uses_provider_token_defaults_and_overrides() -> None:
+    assert get_max_tokens("atlascloud", "qwen/qwen3.5-flash") == 8192
+    assert get_max_tokens("atlas", "deepseek-ai/deepseek-v4-pro", max_output_tokens=2048) == 2048
+
+
+def test_atlas_cloud_openai_compatible_payload_keeps_one_leading_system_message() -> None:
+    client = create_llm_client(
+        provider="atlas-cloud",
+        api_key="test-key",
+        model="deepseek-ai/deepseek-v4-pro",
+    )
+
+    payload = client._build_payload(
+        [
+            LLMMessage(role="user", content="Earlier user turn"),
+            LLMMessage(role="system", content="Static prompt", dynamic_content="Dynamic context"),
+            LLMMessage(role="system", content="Later system note"),
+            LLMMessage(role="user", content="Current user turn"),
+        ],
+        tools=[{"type": "function", "function": {"name": "finish", "parameters": {"type": "object"}}}],
+        temperature=0.2,
+        max_tokens=1024,
+    )
+
+    assert isinstance(client, OpenAICompatibleClient)
+    assert payload["model"] == "deepseek-ai/deepseek-v4-pro"
+    assert payload["tool_choice"] == "auto"
+    assert [message["role"] for message in payload["messages"]].count("system") == 1
+    assert payload["messages"][0]["role"] == "system"
+    assert "Static prompt" in payload["messages"][0]["content"]
+    assert "Dynamic context" in payload["messages"][0]["content"]
+    assert "Later system note" in payload["messages"][0]["content"]

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

@@ -49,6 +49,7 @@ interface RuntimeModelSettings {
 const FALLBACK_LLM_PROVIDERS: LLMProviderSpec[] = [
     { provider: 'anthropic', display_name: 'Anthropic', protocol: 'anthropic', default_base_url: 'https://api.anthropic.com', supports_tool_choice: false, default_max_tokens: 8192 },
     { provider: 'openai', display_name: 'OpenAI', protocol: 'openai_compatible', default_base_url: 'https://api.openai.com/v1', supports_tool_choice: true, default_max_tokens: 16384 },
+    { provider: 'atlascloud', display_name: 'Atlas Cloud', protocol: 'openai_compatible', default_base_url: 'https://api.atlascloud.ai/v1', supports_tool_choice: true, default_max_tokens: 8192 },
     { provider: 'azure', display_name: 'Azure OpenAI', protocol: 'openai_compatible', default_base_url: '', supports_tool_choice: true, default_max_tokens: 16384 },
     { provider: 'deepseek', display_name: 'DeepSeek', protocol: 'openai_compatible', default_base_url: 'https://api.deepseek.com/v1', supports_tool_choice: true, default_max_tokens: 8192 },
     { provider: 'minimax', display_name: 'MiniMax', protocol: 'openai_compatible', default_base_url: 'https://api.minimaxi.com/v1', supports_tool_choice: true, default_max_tokens: 16384 },