feat: add EvoLink LLM provider

backend/.env.sample

@@ -17,6 +17,9 @@ SQLITE_DB_PATH=
 
 # LLM Integrations
 OPENAI_API_KEY=
+EVOLINK_API_KEY=
+EVOLINK_API_BASE=https://direct.evolink.ai/v1
+EVOLINK_CHAT_MODEL=gpt-5.2
 ANTHROPIC_API_KEY=
 GOOGLEAI_API_KEY=
 
@@ -60,4 +63,4 @@ SERP_API_KEY=
 
 # Devzery API Test
 DEVZERY_API_KEY=
-DEVZERY_SOURCE_NAME=
+DEVZERY_SOURCE_NAME=

backend/README.md

@@ -61,7 +61,7 @@ The `VideoDBTool` (`director/tools/videodb_tool.py`) provides an interface for i
 
 ## 🧠 LLM Integration
 
-The `BaseLLM` class (`director/llm/base.py`) provides an abstraction for integrating different Language Model providers. The current implementation supports OpenAI (`director/llm/openai.py`).
+The `BaseLLM` class (`director/llm/base.py`) provides an abstraction for integrating different Language Model providers. The current implementation supports OpenAI (`director/llm/openai.py`) and EvoLink (`director/llm/evolink.py`).
 
 ## 📦 Dependencies
 
@@ -123,4 +123,3 @@ make run
 ## 📚 Further Documentation
 
 For more detailed information about specific components, please refer to the MkDocs documentation.
-

backend/director/constants.py

@@ -18,6 +18,7 @@ class LLMType(str, Enum):
     """Enum for LLM types"""
 
     OPENAI = "openai"
+    EVOLINK = "evolink"
     ANTHROPIC = "anthropic"
     GOOGLEAI = "googleai"
     VIDEODB_PROXY = "videodb_proxy"
@@ -27,7 +28,9 @@ class EnvPrefix(str, Enum):
     """Enum for environment prefixes"""
 
     OPENAI_ = "OPENAI_"
+    EVOLINK_ = "EVOLINK_"
     ANTHROPIC_ = "ANTHROPIC_"
     GOOGLEAI_ = "GOOGLEAI_"
 
-DOWNLOADS_PATH="director/downloads"
+
+DOWNLOADS_PATH = "director/downloads"

backend/director/llm/__init__.py

@@ -3,6 +3,7 @@
 from director.constants import LLMType
 
 from director.llm.openai import OpenAI
+from director.llm.evolink import Evolink
 from director.llm.anthropic import AnthropicAI
 from director.llm.googleai import GoogleAI
 from director.llm.videodb_proxy import VideoDBProxy
@@ -12,13 +13,18 @@ def get_default_llm():
     """Get default LLM"""
 
     openai = True if os.getenv("OPENAI_API_KEY") else False
+    evolink = True if os.getenv("EVOLINK_API_KEY") else False
     anthropic = True if os.getenv("ANTHROPIC_API_KEY") else False
     googleai = True if os.getenv("GOOGLEAI_API_KEY") else False
 
     default_llm = os.getenv("DEFAULT_LLM")
 
-    if openai or default_llm == LLMType.OPENAI:
+    if default_llm == LLMType.EVOLINK:
+        return Evolink()
+    elif openai or default_llm == LLMType.OPENAI:
         return OpenAI()
+    elif evolink:
+        return Evolink()
     elif anthropic or default_llm == LLMType.ANTHROPIC:
         return AnthropicAI()
     elif googleai or default_llm == LLMType.GOOGLEAI:

backend/director/llm/evolink.py

@@ -0,0 +1,52 @@
+from enum import Enum
+
+from pydantic import Field, FieldValidationInfo, field_validator
+from pydantic_settings import SettingsConfigDict
+
+from director.constants import EnvPrefix, LLMType
+from director.llm.openai import OpenAI, OpenaiConfig
+
+
+class EvolinkChatModel(str, Enum):
+    """Enum for EvoLink chat models."""
+
+    GPT5_2 = "gpt-5.2"
+    GPT5_1 = "gpt-5.1"
+    GEMINI_3_1_PRO = "gemini-3.1-pro"
+    DEEPSEEK_V4_PRO = "deepseek-v4-pro"
+    DOUBAO_SEED_2_0_PRO = "doubao-seed-2.0-pro"
+
+
+class EvolinkConfig(OpenaiConfig):
+    """EvoLink Config."""
+
+    model_config = SettingsConfigDict(
+        env_prefix=EnvPrefix.EVOLINK_,
+        extra="ignore",
+    )
+
+    llm_type: str = LLMType.EVOLINK
+    api_key: str = ""
+    api_base: str = "https://direct.evolink.ai/v1"
+    chat_model: str = Field(default=EvolinkChatModel.GPT5_2)
+
+    @field_validator("api_key")
+    @classmethod
+    def validate_non_empty(cls, v, info: FieldValidationInfo):
+        if not v:
+            raise ValueError(
+                f"{info.field_name} must not be empty. Please set {EnvPrefix.EVOLINK_.value}{info.field_name.upper()} environment variable."
+            )
+        return v
+
+
+class Evolink(OpenAI):
+    """EvoLink LLM integration using the OpenAI-compatible Chat Completions API."""
+
+    def __init__(self, config: EvolinkConfig = None):
+        """
+        :param config: EvoLink Config
+        """
+        if config is None:
+            config = EvolinkConfig()
+        super().__init__(config=config)

backend/tests/llm/test_evolink.py

@@ -0,0 +1,42 @@
+import pytest
+
+from director.constants import LLMType
+from director.llm.evolink import Evolink, EvolinkConfig, EvolinkChatModel
+
+
+def test_evolink_config_uses_direct_api_defaults(monkeypatch):
+    monkeypatch.setenv("EVOLINK_API_KEY", "test-key")
+
+    config = EvolinkConfig()
+
+    assert config.llm_type == LLMType.EVOLINK
+    assert config.api_key == "test-key"
+    assert config.api_base == "https://direct.evolink.ai/v1"
+    assert config.chat_model == EvolinkChatModel.GPT5_2
+
+
+def test_evolink_config_supports_env_overrides(monkeypatch):
+    monkeypatch.setenv("EVOLINK_API_KEY", "test-key")
+    monkeypatch.setenv("EVOLINK_API_BASE", "https://example.com/v1")
+    monkeypatch.setenv("EVOLINK_CHAT_MODEL", "deepseek-v4-pro")
+
+    config = EvolinkConfig()
+
+    assert config.api_base == "https://example.com/v1"
+    assert config.chat_model == "deepseek-v4-pro"
+
+
+def test_evolink_config_requires_api_key(monkeypatch):
+    monkeypatch.delenv("EVOLINK_API_KEY", raising=False)
+
+    with pytest.raises(ValueError, match="EVOLINK_API_KEY"):
+        EvolinkConfig()
+
+
+def test_default_llm_can_select_evolink(monkeypatch):
+    import director.llm as llm_module
+
+    monkeypatch.setenv("DEFAULT_LLM", LLMType.EVOLINK)
+    monkeypatch.setenv("EVOLINK_API_KEY", "test-key")
+
+    assert isinstance(llm_module.get_default_llm(), Evolink)

docs/llm/evolink.md

@@ -0,0 +1,17 @@
+## EvoLink
+
+EvoLink extends the OpenAI LLM integration and uses EvoLink's OpenAI-compatible Chat Completions API.
+
+Set `EVOLINK_API_KEY` to your EvoLink API key. The default API base is `https://direct.evolink.ai/v1`, and the default chat model is `gpt-5.2`.
+
+### EvoLink Config
+
+EvoLink Config is the configuration object for EvoLink. It is used to configure EvoLink and is passed to EvoLink when it is created.
+
+::: director.llm.evolink.EvolinkConfig
+
+### EvoLink Interface
+
+EvoLink is the LLM used by the agents and tools. It is used to generate responses to messages.
+
+::: director.llm.evolink.Evolink

mkdocs.yml

@@ -73,6 +73,7 @@ nav:
     - 'Interface': 'llm/interface.md'
     - Integrations:
       - 'OpenAI': 'llm/openai.md'
+      - 'EvoLink': 'llm/evolink.md'
       - 'AnthropicAI': 'llm/anthropic.md'
       - 'GoogleAI': 'llm/googleai.md'
   - 'Database':