Merge branch 'main' into pr-151

nanobot/providers/litellm_provider.py

@@ -1,6 +1,7 @@
 """LiteLLM provider implementation for multi-provider support."""
 
 import json
+import json_repair
 import os
 from typing import Any
 
@@ -15,7 +16,7 @@ class LiteLLMProvider(LLMProvider):
     """
     LLM provider using LiteLLM for multi-provider support.
     
-    Supports OpenRouter, Anthropic, OpenAI, Gemini, and many other providers through
+    Supports OpenRouter, Anthropic, OpenAI, Gemini, MiniMax, and many other providers through
     a unified interface.  Provider-specific logic is driven by the registry
     (see providers/registry.py) — no if-elif chains needed here.
     """
@@ -125,6 +126,10 @@ class LiteLLMProvider(LLMProvider):
         """
         model = self._resolve_model(model or self.default_model)
         
+        # Clamp max_tokens to at least 1 — negative or zero values cause
+        # LiteLLM to reject the request with "max_tokens must be at least 1".
+        max_tokens = max(1, max_tokens)
+        
         kwargs: dict[str, Any] = {
             "model": model,
             "messages": messages,
@@ -135,6 +140,10 @@ class LiteLLMProvider(LLMProvider):
         # Apply model-specific overrides (e.g. kimi-k2.5 temperature)
         self._apply_model_overrides(model, kwargs)
         
+        # Pass api_key directly — more reliable than env vars alone
+        if self.api_key:
+            kwargs["api_key"] = self.api_key
+        
         # Pass api_base for custom endpoints
         if self.api_base:
             kwargs["api_base"] = self.api_base
@@ -168,10 +177,7 @@ class LiteLLMProvider(LLMProvider):
                 # Parse arguments from JSON string if needed
                 args = tc.function.arguments
                 if isinstance(args, str):
-                    try:
-                        args = json.loads(args)
-                    except json.JSONDecodeError:
-                        args = {"raw": args}
+                    args = json_repair.loads(args)
                 
                 tool_calls.append(ToolCallRequest(
                     id=tc.id,

nanobot/providers/registry.py

@@ -66,6 +66,20 @@ class ProviderSpec:
 
 PROVIDERS: tuple[ProviderSpec, ...] = (
 
+    # === Custom (user-provided OpenAI-compatible endpoint) =================
+    # No auto-detection — only activates when user explicitly configures "custom".
+
+    ProviderSpec(
+        name="custom",
+        keywords=(),
+        env_key="OPENAI_API_KEY",
+        display_name="Custom",
+        litellm_prefix="openai",
+        skip_prefixes=("openai/",),
+        is_gateway=True,
+        strip_model_prefix=True,
+    ),
+
     # === Gateways (detected by api_key / api_base, not model name) =========
     # Gateways can route any model, so they win in fallback.
 
@@ -265,6 +279,25 @@ PROVIDERS: tuple[ProviderSpec, ...] = (
         ),
     ),
 
+    # MiniMax: needs "minimax/" prefix for LiteLLM routing.
+    # Uses OpenAI-compatible API at api.minimax.io/v1.
+    ProviderSpec(
+        name="minimax",
+        keywords=("minimax",),
+        env_key="MINIMAX_API_KEY",
+        display_name="MiniMax",
+        litellm_prefix="minimax",            # MiniMax-M2.1 → minimax/MiniMax-M2.1
+        skip_prefixes=("minimax/", "openrouter/"),
+        env_extras=(),
+        is_gateway=False,
+        is_local=False,
+        detect_by_key_prefix="",
+        detect_by_base_keyword="",
+        default_api_base="https://api.minimax.io/v1",
+        strip_model_prefix=False,
+        model_overrides=(),
+    ),
+
     # === Local deployment (matched by config key, NOT by api_base) =========
 
     # vLLM / any OpenAI-compatible local server.