Skip to content

Use Claude Code, Codex, VSCode, Pi, and OpenCode (and 6 other harnesses) for free (1.3B+ free tokens) from your terminal, app, IDE, or phone, and now from the browser with native browser sessions (multi-harness + multi-model) like OpenClaw (voice supported + ToS friendly)

Python

fixed issues after refactor

Alishahryar1f1efa22a82ba
diff --git a/.env.example b/.env.example
index b5322dff..c5f26bb4 100644
--- a/.env.example
+++ b/.env.example
@@ -1,11 +1,9 @@
# ==================== NVIDIA NIM Config ====================
NVIDIA_NIM_API_KEY="nvapi-..."
-NVIDIA_NIM_BASE_URL="https://integrate.api.nvidia.com/v1"
-# ==================== Model Mapping ====================
-# Claude Sonnet/Opus -> BIG_MODEL, Claude Haiku -> SMALL_MODEL
-BIG_MODEL="moonshotai/kimi-k2-thinking"
-SMALL_MODEL="moonshotai/kimi-k2-thinking"
+# ==================== Model ====================
+# All Claude model requests are mapped to this model
+MODEL="moonshotai/kimi-k2-thinking"
# ==================== Rate Limiting ====================
NVIDIA_NIM_RATE_LIMIT=40 # Requests per window
diff --git a/api/app.py b/api/app.py
index ff86c8a9..ce9e1008 100644
--- a/api/app.py
+++ b/api/app.py
@@ -44,12 +44,20 @@ async def lifespan(app: FastAPI):
from messaging.queue import MessageQueueManager
from cli.manager import CLISessionManager
- # Setup workspace
- workspace = os.path.abspath(settings.claude_workspace)
+ # Setup workspace - CLI should run in allowed_dir
+ workspace = (
+ os.path.abspath(settings.allowed_dir)
+ if settings.allowed_dir
+ else os.getcwd()
+ )
os.makedirs(workspace, exist_ok=True)
- # Initialize CLI manager
- allowed_dirs = [settings.allowed_dir] if settings.allowed_dir else []
+ # Session data stored separately
+ data_path = os.path.abspath(settings.claude_workspace)
+ os.makedirs(data_path, exist_ok=True)
+
+ # Initialize CLI manager with allowed_dir as workspace
+ allowed_dirs = [workspace] if settings.allowed_dir else []
cli_manager = CLISessionManager(
workspace_path=workspace,
api_url="http://localhost:8082/v1",
@@ -59,7 +67,7 @@ async def lifespan(app: FastAPI):
# Initialize session store and queue
session_store = SessionStore(
- storage_path=os.path.join(workspace, "sessions.json")
+ storage_path=os.path.join(data_path, "sessions.json")
)
message_queue = MessageQueueManager()
diff --git a/api/dependencies.py b/api/dependencies.py
index e86e77de..2d5acdf7 100644
--- a/api/dependencies.py
+++ b/api/dependencies.py
@@ -3,7 +3,7 @@
from typing import Optional
from providers.base import ProviderConfig
from providers.nvidia_nim import NvidiaNimProvider
-from config.settings import get_settings
+from config.settings import get_settings, NVIDIA_NIM_BASE_URL
# Global provider instance (singleton)
_provider: Optional[NvidiaNimProvider] = None
@@ -16,7 +16,7 @@ def get_provider() -> NvidiaNimProvider:
settings = get_settings()
config = ProviderConfig(
api_key=settings.nvidia_nim_api_key,
- base_url=settings.nvidia_nim_base_url,
+ base_url=NVIDIA_NIM_BASE_URL, # Use constant, not from settings
rate_limit=settings.nvidia_nim_rate_limit,
rate_window=settings.nvidia_nim_rate_window,
)
diff --git a/api/models.py b/api/models.py
index 9923ab58..612e3177 100644
--- a/api/models.py
+++ b/api/models.py
@@ -103,20 +103,23 @@ class MessagesRequest(BaseModel):
@model_validator(mode="after")
def map_model(self) -> "MessagesRequest":
+ """Map any Claude model name to the configured model."""
settings = get_settings()
if self.original_model is None:
self.original_model = self.model
+ # Strip provider prefixes
clean_v = self.model
for prefix in ["anthropic/", "openai/", "gemini/"]:
if clean_v.startswith(prefix):
clean_v = clean_v[len(prefix) :]
break
- if "haiku" in clean_v.lower():
- self.model = settings.small_model
- elif "sonnet" in clean_v.lower() or "opus" in clean_v.lower():
- self.model = settings.big_model
+ # Map all Claude models to the single configured model
+ if any(
+ name in clean_v.lower() for name in ["haiku", "sonnet", "opus", "claude"]
+ ):
+ self.model = settings.model
if self.model != self.original_model:
logger.debug(f"MODEL MAPPING: '{self.original_model}' -> '{self.model}'")
@@ -135,6 +138,7 @@ class TokenCountRequest(BaseModel):
@field_validator("model")
@classmethod
def validate_model_field(cls, v, info):
+ """Map any Claude model name to the configured model."""
settings = get_settings()
clean_v = v
for prefix in ["anthropic/", "openai/", "gemini/"]:
@@ -142,10 +146,10 @@ class TokenCountRequest(BaseModel):
clean_v = clean_v[len(prefix) :]
break
- if "haiku" in clean_v.lower():
- return settings.small_model
- elif "sonnet" in clean_v.lower() or "opus" in clean_v.lower():
- return settings.big_model
+ if any(
+ name in clean_v.lower() for name in ["haiku", "sonnet", "opus", "claude"]
+ ):
+ return settings.model
return v
diff --git a/api/routes.py b/api/routes.py
index 03c7c96c..b8f28a20 100644
--- a/api/routes.py
+++ b/api/routes.py
@@ -233,8 +233,7 @@ async def root():
return {
"status": "ok",
"provider": "nvidia_nim",
- "big_model": settings.big_model,
- "small_model": settings.small_model,
+ "model": settings.model,
}
diff --git a/config/settings.py b/config/settings.py
index 54747675..5786dcbf 100644
--- a/config/settings.py
+++ b/config/settings.py
@@ -8,17 +8,19 @@ from dotenv import load_dotenv
load_dotenv()
+# Fixed base URL for NVIDIA NIM
+NVIDIA_NIM_BASE_URL = "https://integrate.api.nvidia.com/v1"
+
class Settings(BaseSettings):
"""Application settings loaded from environment variables."""
# ==================== NVIDIA NIM Config ====================
nvidia_nim_api_key: str = ""
- nvidia_nim_base_url: str = "https://integrate.api.nvidia.com/v1"
- # ==================== Model Mapping ====================
- big_model: str = "moonshotai/kimi-k2-thinking"
- small_model: str = "moonshotai/kimi-k2-thinking"
+ # ==================== Model ====================
+ # All Claude model requests are mapped to this single model
+ model: str = "moonshotai/kimi-k2-thinking"
# ==================== Rate Limiting ====================
nvidia_nim_rate_limit: int = 40
diff --git a/tests/conftest.py b/tests/conftest.py
index 6fcc655e..9c6a284f 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -4,8 +4,7 @@ import sys
# Set mock environment BEFORE any imports that use Settings
os.environ.setdefault("NVIDIA_NIM_API_KEY", "test_key")
-os.environ.setdefault("BIG_MODEL", "test-model")
-os.environ.setdefault("SMALL_MODEL", "test-model")
+os.environ.setdefault("MODEL", "test-model")
# Add project root to path
sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), "..")))
diff --git a/tests/test_config.py b/tests/test_config.py
index a63af994..7811ff7e 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -42,10 +42,16 @@ class TestSettings:
settings.nvidia_nim_seed, int
)
- def test_model_mapping_defaults(self):
- """Test model mapping defaults."""
+ def test_model_setting(self):
+ """Test model setting exists and is a string."""
from config.settings import Settings
settings = Settings()
- assert "kimi" in settings.big_model.lower() or settings.big_model != ""
- assert "kimi" in settings.small_model.lower() or settings.small_model != ""
+ assert isinstance(settings.model, str)
+ assert len(settings.model) > 0
+
+ def test_base_url_constant(self):
+ """Test NVIDIA_NIM_BASE_URL is a constant."""
+ from config.settings import NVIDIA_NIM_BASE_URL
+
+ assert NVIDIA_NIM_BASE_URL == "https://integrate.api.nvidia.com/v1"