Skip to main content

​
Overview

The Config API provides a centralized way to manage Strix configuration through environment variables and persistent settings.

​
Config Class

Main configuration manager for Strix.

​
Class Variables

Configuration defaults stored as class attributes:
​
strix_llm
str | None
default:"None"
LLM model name
​
llm_api_key
str | None
default:"None"
API key for LLM provider
​
llm_api_base
str | None
default:"None"
Custom API base URL
​
strix_reasoning_effort
str
default:"'high'"
Reasoning effort: “low”, “medium”, or “high”
​
strix_llm_max_retries
str
default:"'5'"
Maximum retry attempts
​
llm_timeout
str
default:"'300'"
Request timeout in seconds
​
strix_image
str
default:"'ghcr.io/usestrix/strix-sandbox:0.1.12'"
Docker image for sandboxes
​
strix_runtime_backend
str
default:"'docker'"
Runtime backend (currently only “docker”)
​
strix_sandbox_execution_timeout
str
default:"'120'"
Tool execution timeout in seconds
​
strix_telemetry
str
default:"'1'"
Enable/disable telemetry

​
Methods

​
get

@classmethod
def get(cls, name: str) -> str | None
Gets a configuration value from environment or defaults.
​
name
str
required
Configuration key in lowercase (e.g., “strix_llm”)
​
return
str | None
Configuration value or None
Example: 
from strix.config import Config

model = Config.get("strix_llm")
print(f"Using model: {model}")

image = Config.get("strix_image")
print(f"Sandbox image: {image}")

​
config_dir

@classmethod
def config_dir(cls) -> Path
Returns the configuration directory path.
​
return
Path
Path to ~/.strix directory
Example: 
from strix.config import Config

config_path = Config.config_dir()
print(f"Config stored in: {config_path}")

​
config_file

@classmethod
def config_file(cls) -> Path
Returns the configuration file path.
​
return
Path
Path to ~/.strix/cli-config.json

​
load

@classmethod
def load(cls) -> dict[str, Any]
Loads configuration from the config file.
​
return
dict[str, Any]
Configuration dictionary
Example: 
from strix.config import Config

config = Config.load()
print(f"Saved environment variables: {config.get('env', {})}")

​
save

@classmethod
def save(cls, config: dict[str, Any]) -> bool
Saves configuration to the config file.
​
config
dict[str, Any]
required
Configuration dictionary to save
​
return
bool
True if successful, False otherwise
Example: 
from strix.config import Config

config_data = {
    "env": {
        "STRIX_LLM": "claude-3-5-sonnet-20241022",
        "LLM_API_KEY": "sk-ant-..."
    }
}

success = Config.save(config_data)
if success:
    print("Configuration saved")

​
apply_saved

@classmethod
def apply_saved(cls, force: bool = False) -> dict[str, str]
Applies saved configuration to environment variables.
​
force
bool
default:"False"
Force apply even if environment variables are already set
​
return
dict[str, str]
Dictionary of applied environment variables
Example: 
from strix.config import Config

# Apply saved config
applied = Config.apply_saved()
print(f"Applied variables: {list(applied.keys())}")

# Force apply
applied = Config.apply_saved(force=True)

​
save_current

@classmethod
def save_current(cls) -> bool
Saves current environment variables to the config file.
​
return
bool
True if successful
Example: 
import os
from strix.config import Config

# Set environment variables
os.environ["STRIX_LLM"] = "gpt-4o"
os.environ["LLM_API_KEY"] = "sk-..."

# Save to config
Config.save_current()

​
tracked_vars

@classmethod
def tracked_vars(cls) -> list[str]
Returns list of tracked environment variable names.
​
return
list[str]
List of uppercase environment variable names
Example: 
from strix.config import Config

tracked = Config.tracked_vars()
print(f"Tracked variables: {tracked}")
# Output: ['STRIX_LLM', 'LLM_API_KEY', 'LLM_API_BASE', ...]

​
Helper Functions

​
apply_saved_config

from strix.config import apply_saved_config

applied = apply_saved_config(force: bool = False) -> dict[str, str]
Applies saved configuration to environment.
​
force
bool
default:"False"
Force apply variables
​
return
dict[str, str]
Applied variables

​
save_current_config

from strix.config import save_current_config

success = save_current_config() -> bool
Saves current environment to config file.
​
return
bool
True if successful

​
resolve_llm_config

from strix.config import resolve_llm_config

model, api_key, api_base = resolve_llm_config() -> tuple[str | None, str | None, str | None]
Resolves LLM configuration from environment.
​
return
tuple[str | None, str | None, str | None]
Tuple of (model_name, api_key, api_base)
Example: 
from strix.config import resolve_llm_config

model, api_key, api_base = resolve_llm_config()

if model:
    print(f"Model: {model}")
    print(f"API Base: {api_base or 'default'}")
else:
    print("No LLM configured")

​
Configuration File Format

The config file is stored at ~/.strix/cli-config.json: 
{
  "env": {
    "STRIX_LLM": "claude-3-5-sonnet-20241022",
    "LLM_API_KEY": "sk-ant-...",
    "STRIX_IMAGE": "ghcr.io/usestrix/strix-sandbox:0.1.12",
    "STRIX_RUNTIME_BACKEND": "docker"
  }
}

​
Environment Variables

​
LLM Configuration

​
STRIX_LLM
str
required
Model name (e.g., “claude-3-5-sonnet-20241022”)
​
LLM_API_KEY
str
required
API key for the provider
​
LLM_API_BASE
str
default:"None"
Custom API base URL
​
OPENAI_API_BASE
str
default:"None"
OpenAI-specific API base
​
LITELLM_BASE_URL
str
default:"None"
LiteLLM base URL
​
OLLAMA_API_BASE
str
default:"None"
Ollama API base
​
STRIX_REASONING_EFFORT
str
default:"'high'"
Reasoning effort for o1 models: “low”, “medium”, “high”
​
STRIX_LLM_MAX_RETRIES
str
default:"'5'"
Maximum retry attempts
​
STRIX_MEMORY_COMPRESSOR_TIMEOUT
str
default:"'30'"
Memory compression timeout in seconds
​
LLM_TIMEOUT
str
default:"'300'"
Request timeout in seconds

​
Runtime Configuration

​
STRIX_IMAGE
str
default:"'ghcr.io/usestrix/strix-sandbox:0.1.12'"
Docker image for sandboxes
​
STRIX_RUNTIME_BACKEND
str
default:"'docker'"
Runtime backend
​
STRIX_SANDBOX_EXECUTION_TIMEOUT
str
default:"'120'"
Tool execution timeout in seconds
​
STRIX_SANDBOX_CONNECT_TIMEOUT
str
default:"'10'"
Connection timeout in seconds

​
Feature Configuration

​
PERPLEXITY_API_KEY
str
default:"None"
API key for Perplexity web search
​
STRIX_DISABLE_BROWSER
str
default:"'false'"
Disable browser tools
​
STRIX_TELEMETRY
str
default:"'1'"
Enable telemetry (1 or 0)

​
Usage Examples

​
Basic Configuration

import os
from strix.config import Config

# Set environment variables
os.environ["STRIX_LLM"] = "claude-3-5-sonnet-20241022"
os.environ["LLM_API_KEY"] = "sk-ant-..."

# Get configuration values
model = Config.get("strix_llm")
api_key = Config.get("llm_api_key")

print(f"Model: {model}")
print(f"API Key: {api_key[:10]}...")

​
Persistent Configuration

import os
from strix.config import Config

# Set configuration
os.environ["STRIX_LLM"] = "gpt-4o"
os.environ["LLM_API_KEY"] = "sk-..."
os.environ["STRIX_IMAGE"] = "ghcr.io/usestrix/strix-sandbox:latest"

# Save to file
Config.save_current()
print("Configuration saved to ~/.strix/cli-config.json")

# Later, in another session
from strix.config import apply_saved_config

# Load saved configuration
applied = apply_saved_config()
print(f"Applied {len(applied)} variables from saved config")

​
Custom Configuration File

from pathlib import Path
from strix.config import Config

# Use custom config file
Config._config_file_override = Path("/custom/path/config.json")

# Load from custom location
config = Config.load()
print(f"Loaded config from custom path: {config}")

​
Checking Configuration

from strix.config import Config, resolve_llm_config

# Check what's configured
model, api_key, api_base = resolve_llm_config()

if not model:
    print("ERROR: STRIX_LLM not configured")
else:
    print(f"Model: {model}")
    print(f"API Base: {api_base or 'default'}")
    print(f"Has API Key: {bool(api_key)}")

# List all tracked variables
tracked = Config.tracked_vars()
print(f"\nTracked variables ({len(tracked)}):")
for var in tracked:
    value = os.environ.get(var)
    if value:
        display = value[:20] + "..." if len(value) > 20 else value
        print(f"  {var}: {display}")

Build docs developers (and LLMs) love

Get started for freeTalk to us