Skip to main content

config Command

Manage FluxLoop configuration settings.

Overview​

The config command provides utilities for viewing, updating, and validating FluxLoop configuration files. It supports both reading/writing YAML configs and managing environment variables.

Subcommands​

show​

Display configuration file contents.

# Show project configuration
fluxloop config show --file configs/project.yaml

# Show in JSON format
fluxloop config show --file configs/simulation.yaml --format json

# Show with syntax highlighting and line numbers
fluxloop config show

Options:

OptionDescriptionDefault
--file, -fConfiguration file to showconfigs/project.yaml
--projectProject name under FluxLoop rootCurrent directory
--rootFluxLoop root directory./fluxloop
--formatOutput format (yaml or json)yaml

Example:

# View simulation config
fluxloop config show --file configs/simulation.yaml

# View as JSON for parsing
fluxloop config show --format json

set​

Update a specific configuration value.

# Set a simple value
fluxloop config set iterations 20

# Set nested value with dot notation
fluxloop config set runner.timeout_seconds 300

# Set boolean value
fluxloop config set multi_turn.enabled true

Arguments:

ArgumentDescription
keyConfiguration key to set (use dot notation for nested keys)
valueValue to set (auto-detects type: int, float, bool, or string)

Options:

OptionDescriptionDefault
--file, -fConfiguration file to updateconfigs/project.yaml
--projectProject nameCurrent
--rootFluxLoop root directory./fluxloop

Type Detection:

The command automatically detects value types:

  • Integer: 20, 100
  • Float: 0.5, 3.14
  • Boolean: true, false (case-insensitive)
  • String: Everything else

Examples:

# Update iteration count
fluxloop config set iterations 50 --file configs/simulation.yaml

# Set timeout in seconds
fluxloop config set runner.timeout_seconds 180

# Enable multi-turn mode
fluxloop config set multi_turn.enabled true

# Set temperature for variation
fluxloop config set variation_temperature 0.9 --file configs/input.yaml

# Set runner target
fluxloop config set runner.target "my_agent:run"

env​

Display environment variables used by FluxLoop.

# Show environment variables (masks API keys)
fluxloop config env

# Show all values including sensitive keys
fluxloop config env --show-values

Options:

OptionDescriptionDefault
--show-values, -sShow actual values (including secrets)false
--projectProject nameCurrent
--rootFluxLoop root directory./fluxloop

Environment Variables Shown:

VariableDescriptionDefault
FLUXLOOP_COLLECTOR_URLCollector service URLhttp://localhost:8000
FLUXLOOP_API_KEYAPI key for authenticationNone
FLUXLOOP_ENABLEDEnable/disable tracingtrue
FLUXLOOP_DEBUGEnable debug modefalse
FLUXLOOP_SAMPLE_RATETrace sampling rate (0-1)1.0
FLUXLOOP_SERVICE_NAMEService name for tracesNone
FLUXLOOP_ENVIRONMENTEnvironment (dev/staging/prod)development
OPENAI_API_KEYOpenAI API keyNone
ANTHROPIC_API_KEYAnthropic API keyNone

Output:

╭─ FluxLoop Environment Variables ──────────────────────────────────╮
│ Variable                  Description             Current Value   │
│ FLUXLOOP_COLLECTOR_URL    Collector service URL   Not set        │
│ FLUXLOOP_ENABLED          Enable/disable tracing  true           │
│ OPENAI_API_KEY            OpenAI API key          ****abc123     │
╰────────────────────────────────────────────────────────────────────╯

Loaded from: /path/to/.env

Notes:

  • API keys are masked by default (shows last 4 characters)
  • Use --show-values to see full keys (be careful with logs!)
  • Environment files are merged: root .env → project .env

set-llm​

Configure LLM provider credentials and defaults.

# Set OpenAI as LLM provider
fluxloop config set-llm openai sk-xxxxx

# Set OpenAI with specific model
fluxloop config set-llm openai sk-xxxxx --model gpt-4o

# Set Anthropic provider
fluxloop config set-llm anthropic sk-ant-xxxxx --model claude-3-5-sonnet-20241022

# Overwrite existing API key
fluxloop config set-llm openai sk-newkey --overwrite-env

Arguments:

ArgumentDescription
providerLLM provider (openai, anthropic, gemini)
api_keyAPI key or token for the provider

Options:

OptionDescriptionDefault
--model, -mDefault model to useProvider default
--overwrite-envOverwrite existing key in .envfalse
--file, -fConfig file to updateconfigs/input.yaml
--env-filePath to environment file.env
--projectProject nameCurrent
--rootFluxLoop root directory./fluxloop

Supported Providers:

ProviderEnvironment VariableDefault Model
openaiOPENAI_API_KEYgpt-5
anthropicANTHROPIC_API_KEYclaude-3-haiku-20240307
geminiGEMINI_API_KEYgemini-2.5-flash

What This Command Does:

  1. Saves API key to .env:

    OPENAI_API_KEY=sk-xxxxx

  2. Updates configs/input.yaml:

    input_generation:
      llm:
        enabled: true
        provider: openai
        model: gpt-4o
        api_key: null  # References OPENAI_API_KEY from .env

Examples:

# Quick setup for OpenAI
fluxloop config set-llm openai sk-proj-abc123

# Use specific Anthropic model
fluxloop config set-llm anthropic sk-ant-xyz789 \
  --model claude-3-5-sonnet-20241022

# Update existing key
fluxloop config set-llm openai sk-new-key --overwrite-env

# Set for specific project
fluxloop config set-llm openai sk-key --project my-agent

validate​

Validate configuration file structure and values.

# Validate default project config
fluxloop config validate

# Validate specific config file
fluxloop config validate --file configs/simulation.yaml

# Validate for specific project
fluxloop config validate --project my-agent

Options:

OptionDescriptionDefault
--file, -fConfiguration file to validateconfigs/project.yaml
--projectProject nameCurrent
--rootFluxLoop root directory./fluxloop

What Gets Validated:

  • YAML syntax correctness
  • Required fields present
  • Field types match schema
  • Runner configuration valid
  • Persona definitions valid
  • Evaluator configurations valid

Success Output:

✓ Configuration is valid!

╭──────────────────────────────────────╮
│ Property      │ Value                │
│ Experiment    │ my_agent_experiment  │
│ Iterations    │ 10                   │
│ Personas      │ 2                    │
│ Variations    │ 5                    │
│ Total Runs    │ 100                  │
│ Runner Module │ examples.simple_agent│
│ Evaluators    │ 3                    │
╰──────────────────────────────────────╯

Error Output:

✗ Validation failed: Missing required field: runner.module_path

Common Validation Errors:

ErrorCauseSolution
Missing required fieldConfig missing mandatory keyAdd the required field
Invalid YAML syntaxMalformed YAMLCheck indentation and quotes
Type mismatchWrong value typeUse correct type (int, str, etc.)
Runner module not foundModule path incorrectVerify module exists
Invalid personaMalformed persona definitionCheck persona structure

Common Workflows​

Initial Setup​

# 1. Initialize project
fluxloop init project --name my-agent

# 2. Set LLM provider
fluxloop config set-llm openai sk-xxxxx --model gpt-4o

# 3. Validate configuration
fluxloop config validate

# 4. Check environment
fluxloop config env

Tuning Experiment Settings​

# Increase iterations for more samples
fluxloop config set iterations 50 --file configs/simulation.yaml

# Adjust variation temperature
fluxloop config set variation_temperature 0.8 --file configs/input.yaml

# Enable multi-turn mode
fluxloop config set multi_turn.enabled true --file configs/simulation.yaml
fluxloop config set multi_turn.max_turns 10 --file configs/simulation.yaml

Switching LLM Providers​

# Switch from OpenAI to Anthropic
fluxloop config set-llm anthropic sk-ant-xxxxx \
  --model claude-3-5-sonnet-20241022 \
  --overwrite-env

Verifying Configuration Before Running​

# Full validation check
fluxloop config validate --file configs/simulation.yaml
fluxloop config validate --file configs/input.yaml
fluxloop config validate --file configs/evaluation.yaml

# Check environment setup
fluxloop config env

# Or use doctor command for complete diagnostics
fluxloop doctor

Configuration File Locations​

Default configuration structure:

fluxloop/my-agent/
├── configs/
│   ├── project.yaml       # Project metadata
│   ├── input.yaml         # Personas and input generation
│   ├── simulation.yaml    # Runner and experiment settings
│   └── evaluation.yaml    # Evaluators and success criteria
├── .env                   # Environment variables (gitignored)
├── inputs/
│   └── generated.yaml
├── recordings/
└── experiments/

Tips​

Dot Notation for Nested Keys​

Use dot notation to set deeply nested values:

# Set nested runner config
fluxloop config set runner.timeout_seconds 300
fluxloop config set runner.max_retries 5

# Set multi-turn supervisor settings
fluxloop config set multi_turn.supervisor.model gpt-4o
fluxloop config set multi_turn.supervisor.provider openai

Viewing Before Modifying​

Always check current values before changing:

# Show current config
fluxloop config show --file configs/simulation.yaml

# Make change
fluxloop config set iterations 20 --file configs/simulation.yaml

# Verify change
fluxloop config show --file configs/simulation.yaml

Project-Specific vs Global​

Use --project flag to manage multiple projects:

# Set config for specific project
fluxloop config set iterations 100 --project production-agent

# Different config for dev project
fluxloop config set iterations 10 --project dev-agent

See Also​