Merge branch 'main' of https://github.com/HKUDS/nanobot

README.md

@@ -574,7 +574,7 @@ Config file: `~/.nanobot/config.json`
 
 | Provider | Purpose | Get API Key |
 |----------|---------|-------------|
-| `custom` | Any OpenAI-compatible endpoint | — |
+| `custom` | Any OpenAI-compatible endpoint (direct, no LiteLLM) | — |
 | `openrouter` | LLM (recommended, access to all models) | [openrouter.ai](https://openrouter.ai) |
 | `anthropic` | LLM (Claude direct) | [console.anthropic.com](https://console.anthropic.com) |
 | `openai` | LLM (GPT direct) | [platform.openai.com](https://platform.openai.com) |
@@ -623,7 +623,7 @@ nanobot agent -m "Hello!"
 <details>
 <summary><b>Custom Provider (Any OpenAI-compatible API)</b></summary>
 
-If your provider is not listed above but exposes an **OpenAI-compatible API** (e.g. Together AI, Fireworks, Azure OpenAI, self-hosted endpoints), use the `custom` provider:
+Connects directly to any OpenAI-compatible endpoint — LM Studio, llama.cpp, Together AI, Fireworks, Azure OpenAI, or any self-hosted server. Bypasses LiteLLM; model name is passed as-is.
 
 ```json
 {
@@ -641,7 +641,7 @@ If your provider is not listed above but exposes an **OpenAI-compatible API** (e
 }
 ```
 
-> The `custom` provider routes through LiteLLM's OpenAI-compatible path. It works with any endpoint that follows the OpenAI chat completions API format. The model name is passed directly to the endpoint without any prefix.
+> For local servers that don't require a key, set `apiKey` to any non-empty string (e.g. `"no-key"`).
 
 </details>
 

nanobot/cli/commands.py

@@ -280,18 +280,27 @@ This file stores important information that should persist across sessions.
 
 
 def _make_provider(config: Config):
-    """Create LiteLLMProvider from config. Exits if no API key found."""
+    """Create the appropriate LLM provider from config."""
     from nanobot.providers.litellm_provider import LiteLLMProvider
     from nanobot.providers.openai_codex_provider import OpenAICodexProvider
+    from nanobot.providers.custom_provider import CustomProvider
 
     model = config.agents.defaults.model
     provider_name = config.get_provider_name(model)
     p = config.get_provider(model)
 
-    # OpenAI Codex (OAuth): don't route via LiteLLM; use the dedicated implementation.
+    # OpenAI Codex (OAuth)
     if provider_name == "openai_codex" or model.startswith("openai-codex/"):
         return OpenAICodexProvider(default_model=model)
 
+    # Custom: direct OpenAI-compatible endpoint, bypasses LiteLLM
+    if provider_name == "custom":
+        return CustomProvider(
+            api_key=p.api_key if p else "no-key",
+            api_base=config.get_api_base(model) or "http://localhost:8000/v1",
+            default_model=model,
+        )
+
     from nanobot.providers.registry import find_by_name
     spec = find_by_name(provider_name)
     if not model.startswith("bedrock/") and not (p and p.api_key) and not (spec and spec.is_oauth):

nanobot/providers/custom_provider.py

@@ -0,0 +1,47 @@
+"""Direct OpenAI-compatible provider — bypasses LiteLLM."""
+
+from __future__ import annotations
+
+from typing import Any
+
+import json_repair
+from openai import AsyncOpenAI
+
+from nanobot.providers.base import LLMProvider, LLMResponse, ToolCallRequest
+
+
+class CustomProvider(LLMProvider):
+
+    def __init__(self, api_key: str = "no-key", api_base: str = "http://localhost:8000/v1", default_model: str = "default"):
+        super().__init__(api_key, api_base)
+        self.default_model = default_model
+        self._client = AsyncOpenAI(api_key=api_key, base_url=api_base)
+
+    async def chat(self, messages: list[dict[str, Any]], tools: list[dict[str, Any]] | None = None,
+                   model: str | None = None, max_tokens: int = 4096, temperature: float = 0.7) -> LLMResponse:
+        kwargs: dict[str, Any] = {"model": model or self.default_model, "messages": messages,
+                                  "max_tokens": max(1, max_tokens), "temperature": temperature}
+        if tools:
+            kwargs.update(tools=tools, tool_choice="auto")
+        try:
+            return self._parse(await self._client.chat.completions.create(**kwargs))
+        except Exception as e:
+            return LLMResponse(content=f"Error: {e}", finish_reason="error")
+
+    def _parse(self, response: Any) -> LLMResponse:
+        choice = response.choices[0]
+        msg = choice.message
+        tool_calls = [
+            ToolCallRequest(id=tc.id, name=tc.function.name,
+                            arguments=json_repair.loads(tc.function.arguments) if isinstance(tc.function.arguments, str) else tc.function.arguments)
+            for tc in (msg.tool_calls or [])
+        ]
+        u = response.usage
+        return LLMResponse(
+            content=msg.content, tool_calls=tool_calls, finish_reason=choice.finish_reason or "stop",
+            usage={"prompt_tokens": u.prompt_tokens, "completion_tokens": u.completion_tokens, "total_tokens": u.total_tokens} if u else {},
+            reasoning_content=getattr(msg, "reasoning_content", None),
+        )
+
+    def get_default_model(self) -> str:
+        return self.default_model

nanobot/providers/registry.py

@@ -54,6 +54,9 @@ class ProviderSpec:
     # OAuth-based providers (e.g., OpenAI Codex) don't use API keys
     is_oauth: bool = False                   # if True, uses OAuth flow instead of API key
 
+    # Direct providers bypass LiteLLM entirely (e.g., CustomProvider)
+    is_direct: bool = False
+
     @property
     def label(self) -> str:
         return self.display_name or self.name.title()
@@ -65,18 +68,14 @@ class ProviderSpec:
 
 PROVIDERS: tuple[ProviderSpec, ...] = (
 
-    # === Custom (user-provided OpenAI-compatible endpoint) =================
-    # No auto-detection — only activates when user explicitly configures "custom".
-
+    # === Custom (direct OpenAI-compatible endpoint, bypasses LiteLLM) ======
     ProviderSpec(
         name="custom",
         keywords=(),
-        env_key="OPENAI_API_KEY",
+        env_key="",
         display_name="Custom",
-        litellm_prefix="openai",
-        skip_prefixes=("openai/",),
-        is_gateway=True,
-        strip_model_prefix=True,
+        litellm_prefix="",
+        is_direct=True,
     ),
 
     # === Gateways (detected by api_key / api_base, not model name) =========