Configuration

Complete reference for all YOLO Developer configuration options.

Table of contents

  1. Overview
  2. Configuration File
  3. Configuration Sections
    1. project_name
    2. llm
      1. llm.openai
      2. llm.hybrid
      3. Supported Models
    3. quality
      1. Available Gates
    4. memory
      1. Vector Store Types
      2. Graph Store Types
    5. brownfield
    6. analyst
    7. web
    8. github
      1. github.import_config
      2. github.import_config.story
    9. agents
      1. Per-Agent Configuration
    10. orchestrator
    11. audit
    12. mcp
    13. tools
      1. Benefits of External Tool Integration
      2. Checking Tool Status
  4. Environment Variables
    1. Syntax
    2. Examples
    3. List Values
  5. Configuration Priority
    1. Example
  6. Configuration Validation
    1. Validate Configuration
    2. Common Validation Errors
  7. Configuration Profiles
    1. Development Profile
    2. Production Profile
    3. Using Profiles
  8. Complete Example
  9. Next Steps

Overview

YOLO Developer uses a three-layer configuration system:

  1. Defaults - Built-in sensible defaults
  2. YAML file - Project-specific yolo.yaml
  3. Environment variables - Runtime overrides (highest priority)

Configuration File

Create yolo.yaml in your project root:

# Project identification
project_name: my-awesome-api

# LLM configuration
llm:
  provider: auto
  cheap_model: gpt-5.2-instant
  premium_model: claude-sonnet-4-20250514
  best_model: claude-opus-4-5-20251101
  openai:
    code_model: gpt-5.2-pro
  # API keys should be set via environment variables

# Quality thresholds
quality:
  test_coverage_threshold: 0.8
  gate_pass_threshold: 0.7
  blocking_gates:
    - testability
    - architecture_validation
    - definition_of_done

# Memory and storage
memory:
  vector_store_type: chromadb
  graph_store_type: json
  persist_path: .yolo/memory

# Agent configuration
agents:
  max_iterations: 10
  timeout_seconds: 300
  enable_caching: true

# Orchestration
orchestrator:
  max_agent_retries: 3
  circular_detection_threshold: 5
  human_escalation_enabled: true

# Audit and logging
audit:
  enabled: true
  log_level: info
  export_format: json
  retention_days: 30

# MCP server
mcp:
  transport: stdio
  http:
    port: 8080
    host: "127.0.0.1"

Configuration Sections

project_name

Type: string Required: No (defaults to directory name) Environment: YOLO_PROJECT_NAME

The name of your project. Used in logs, audit trail, and generated artifacts.

project_name: my-awesome-api

llm

LLM (Large Language Model) configuration for agent reasoning.

Option Type Default Env Var Description
provider string auto YOLO_LLM__PROVIDER Primary provider (auto/openai/anthropic/hybrid)
cheap_model string gpt-5.2-instant YOLO_LLM__CHEAP_MODEL Model for routine tasks
premium_model string claude-sonnet-4-20250514 YOLO_LLM__PREMIUM_MODEL Model for complex reasoning
best_model string claude-opus-4-5-20251101 YOLO_LLM__BEST_MODEL Model for critical decisions
openai_api_key string None YOLO_LLM__OPENAI_API_KEY OpenAI API key (legacy)
anthropic_api_key string None YOLO_LLM__ANTHROPIC_API_KEY Anthropic API key

Never put API keys in yolo.yaml. Always use environment variables. For local development only, set YOLO_ALLOW_YAML_SECRETS=1 to allow keys in yolo.yaml (do not commit secrets).

llm:
  provider: auto
  cheap_model: gpt-5.2-instant
  premium_model: claude-sonnet-4-20250514
  best_model: claude-opus-4-5-20251101
  openai:
    cheap_model: gpt-5.2-instant
    premium_model: gpt-5.2-thinking
    code_model: gpt-5.2-pro
    reasoning_model: gpt-5.2-pro
  hybrid:
    enabled: false
    routing:
      code_generation: openai
      code_review: openai
      architecture: anthropic
      analysis: anthropic
      documentation: openai
      testing: openai

Environment Variables:

export YOLO_LLM__OPENAI__API_KEY=sk-proj-...
export YOLO_LLM__OPENAI_API_KEY=sk-proj-...  # Legacy
export YOLO_LLM__ANTHROPIC_API_KEY=sk-ant-...
export YOLO_LLM__PREMIUM_MODEL=claude-sonnet-4-20250514

llm.openai

OpenAI/Codex configuration for code-optimized models.

Option Type Default Env Var Description
api_key string None YOLO_LLM__OPENAI__API_KEY OpenAI API key (preferred)
cheap_model string gpt-5.2-instant YOLO_LLM__OPENAI__CHEAP_MODEL OpenAI model for routine tasks
premium_model string gpt-5.2-thinking YOLO_LLM__OPENAI__PREMIUM_MODEL OpenAI model for complex reasoning
code_model string gpt-5.2-pro YOLO_LLM__OPENAI__CODE_MODEL OpenAI model for code tasks
reasoning_model string gpt-5.2-pro YOLO_LLM__OPENAI__REASONING_MODEL OpenAI model for deep reasoning

llm.hybrid

Hybrid routing configuration for task-based provider selection.

Option Type Default Env Var Description
enabled bool false YOLO_LLM__HYBRID__ENABLED Enable hybrid routing
routing.code_generation string openai YOLO_LLM__HYBRID__ROUTING__CODE_GENERATION Provider for code generation
routing.code_review string openai YOLO_LLM__HYBRID__ROUTING__CODE_REVIEW Provider for code review
routing.architecture string anthropic YOLO_LLM__HYBRID__ROUTING__ARCHITECTURE Provider for architecture
routing.analysis string anthropic YOLO_LLM__HYBRID__ROUTING__ANALYSIS Provider for analysis
routing.documentation string openai YOLO_LLM__HYBRID__ROUTING__DOCUMENTATION Provider for documentation
routing.testing string openai YOLO_LLM__HYBRID__ROUTING__TESTING Provider for testing

Supported Models

OpenAI:

  • gpt-5.2-pro (recommended for code + deep reasoning)
  • gpt-5.2-thinking (recommended for premium reasoning)
  • gpt-5.2-instant (recommended for cheap/fast)

Anthropic:

  • claude-3-opus-20240229
  • claude-3-sonnet-20240229
  • claude-3-haiku-20240307

quality

Quality gate thresholds and configuration.

Option Type Default Env Var Description
test_coverage_threshold float 0.8 YOLO_QUALITY__TEST_COVERAGE_THRESHOLD Minimum test coverage (0.0-1.0)
gate_pass_threshold float 0.7 YOLO_QUALITY__GATE_PASS_THRESHOLD Minimum gate score (0.0-1.0)
blocking_gates list [testability, architecture_validation] YOLO_QUALITY__BLOCKING_GATES Gates that block progress
warning_gates list [] YOLO_QUALITY__WARNING_GATES Gates that only warn 
quality:
  test_coverage_threshold: 0.8
  gate_pass_threshold: 0.7
  blocking_gates:
    - testability
    - ac_measurability
    - architecture_validation
    - definition_of_done
  warning_gates:
    - code_complexity

Available Gates

Gate Description
testability Validates stories have testable acceptance criteria
ac_measurability Checks acceptance criteria are measurable
architecture_validation Validates design against patterns and constraints
definition_of_done Checks completed work meets DoD criteria
code_complexity Analyzes cyclomatic complexity
security_scan Runs security checks on generated code

memory

Memory and vector storage configuration.

Option Type Default Env Var Description
persist_path string .yolo/memory YOLO_MEMORY__PERSIST_PATH Storage location
vector_store_type string chromadb YOLO_MEMORY__VECTOR_STORE_TYPE Vector store backend
graph_store_type string json YOLO_MEMORY__GRAPH_STORE_TYPE Graph store backend 
memory:
  persist_path: .yolo/memory
  vector_store_type: chromadb
  graph_store_type: json

Vector Store Types

Backend Description
chromadb Default, local ChromaDB instance

Graph Store Types

Backend Description
json Default, local JSON graph store
neo4j Optional Neo4j backend

brownfield

Brownfield scanning configuration for existing projects.

Option Type Default Env Var Description
scan_depth int 3 YOLO_BROWNFIELD__SCAN_DEPTH Directory depth to scan
exclude_patterns list node_modules, .git, pycache, .venv, venv YOLO_BROWNFIELD__EXCLUDE_PATTERNS Patterns to skip during scan
include_git_history bool true YOLO_BROWNFIELD__INCLUDE_GIT_HISTORY Include git metadata
max_files_to_analyze int 1000 YOLO_BROWNFIELD__MAX_FILES_TO_ANALYZE File limit for scanning
interactive bool true YOLO_BROWNFIELD__INTERACTIVE Prompt for ambiguous findings 
brownfield:
  scan_depth: 3
  exclude_patterns:
    - node_modules
    - .git
    - __pycache__
    - .venv
    - venv
  include_git_history: true
  max_files_to_analyze: 1000
  interactive: true

analyst

Interactive requirements gathering configuration.

Option Type Default Env Var Description
gathering.enabled bool true YOLO_ANALYST__GATHERING__ENABLED Enable gathering sessions
gathering.storage_path string .yolo/sessions YOLO_ANALYST__GATHERING__STORAGE_PATH Session storage
gathering.max_questions_per_phase int 5 YOLO_ANALYST__GATHERING__MAX_QUESTIONS_PER_PHASE Question cap 
analyst:
  gathering:
    enabled: true
    storage_path: .yolo/sessions
    max_questions_per_phase: 5

web

Local web UI configuration.

Option Type Default Env Var Description
enabled bool true YOLO_WEB__ENABLED Enable web UI
host string 127.0.0.1 YOLO_WEB__HOST Host to bind
port int 3000 YOLO_WEB__PORT Port to bind
api_only bool false YOLO_WEB__API_ONLY Run API only
uploads.enabled bool true YOLO_WEB__UPLOADS__ENABLED Enable uploads
uploads.max_size_mb int 10 YOLO_WEB__UPLOADS__MAX_SIZE_MB Max upload size
uploads.storage_path string .yolo/uploads YOLO_WEB__UPLOADS__STORAGE_PATH Upload storage 
web:
  enabled: true
  host: 127.0.0.1
  port: 3000
  api_only: false
  uploads:
    enabled: true
    max_size_mb: 10
    storage_path: .yolo/uploads

github

GitHub automation configuration for repository management.

Prerequisite: install the GitHub CLI (gh) and authenticate (gh auth login).

Option Type Default Env Var Description
enabled bool false YOLO_GITHUB__ENABLED Enable GitHub automation
token string None YOLO_GITHUB__TOKEN GitHub token (env only)
repository string None YOLO_GITHUB__REPOSITORY owner/repo override
default_branch string main YOLO_GITHUB__DEFAULT_BRANCH Default branch
branch_prefix string feature/ YOLO_GITHUB__BRANCH_PREFIX Branch prefix 
github:
  enabled: true
  repository: bbengt1/yolo-developer
  default_branch: main
  branch_prefix: feature/
  automation:
    auto_commit: true
    auto_push: true
    auto_pr: true
  pull_requests:
    draft_by_default: false
    merge_method: squash
  issues:
    create_from_stories: true
  releases:
    generate_notes: true
  commits:
    conventional: true
  import_config:
    enabled: true
    update_issues: true
    add_label: yolo-imported
    story:
      id_prefix: US
      include_technical_notes: true
      estimate_points: true

github.import_config

GitHub issue import configuration.

Option Type Default Env Var Description
enabled bool true YOLO_GITHUB__IMPORT_CONFIG__ENABLED Enable issue import
default_repo string None YOLO_GITHUB__IMPORT_CONFIG__DEFAULT_REPO Default repo for imports
update_issues bool true YOLO_GITHUB__IMPORT_CONFIG__UPDATE_ISSUES Post back to issues
add_label string yolo-imported YOLO_GITHUB__IMPORT_CONFIG__ADD_LABEL Label to apply
add_comment bool true YOLO_GITHUB__IMPORT_CONFIG__ADD_COMMENT Comment with story summary

github.import_config.story

Option Type Default Env Var Description
id_prefix string US YOLO_GITHUB__IMPORT_CONFIG__STORY__ID_PREFIX Story ID prefix
include_technical_notes bool true YOLO_GITHUB__IMPORT_CONFIG__STORY__INCLUDE_TECHNICAL_NOTES Include technical notes
estimate_points bool true YOLO_GITHUB__IMPORT_CONFIG__STORY__ESTIMATE_POINTS Estimate points

agents

Agent execution configuration.

Option Type Default Env Var Description
max_iterations int 10 YOLO_AGENTS__MAX_ITERATIONS Max iterations per agent
timeout_seconds int 300 YOLO_AGENTS__TIMEOUT_SECONDS Agent timeout
enable_caching bool true YOLO_AGENTS__ENABLE_CACHING Cache agent responses
cache_ttl_hours int 24 YOLO_AGENTS__CACHE_TTL_HOURS Cache expiry time
parallel_stories int 1 YOLO_AGENTS__PARALLEL_STORIES Stories to process in parallel 
agents:
  max_iterations: 10
  timeout_seconds: 300
  enable_caching: true
  cache_ttl_hours: 24
  parallel_stories: 1

Per-Agent Configuration 

agents:
  max_iterations: 10

  analyst:
    max_iterations: 5
    timeout_seconds: 120

  dev:
    max_iterations: 15
    timeout_seconds: 600
    parallel_stories: 2

  tea:
    timeout_seconds: 180

orchestrator

Orchestration and workflow configuration.

Option Type Default Env Var Description
max_agent_retries int 3 YOLO_ORCHESTRATOR__MAX_AGENT_RETRIES Retries on agent failure
circular_detection_threshold int 5 YOLO_ORCHESTRATOR__CIRCULAR_DETECTION_THRESHOLD Detect circular logic after N loops
human_escalation_enabled bool true YOLO_ORCHESTRATOR__HUMAN_ESCALATION_ENABLED Allow human escalation
checkpoint_enabled bool true YOLO_ORCHESTRATOR__CHECKPOINT_ENABLED Enable checkpoints
checkpoint_interval int 5 YOLO_ORCHESTRATOR__CHECKPOINT_INTERVAL Checkpoint every N operations 
orchestrator:
  max_agent_retries: 3
  circular_detection_threshold: 5
  human_escalation_enabled: true
  checkpoint_enabled: true
  checkpoint_interval: 5

audit

Audit trail and logging configuration.

Option Type Default Env Var Description
enabled bool true YOLO_AUDIT__ENABLED Enable audit logging
log_level string info YOLO_AUDIT__LOG_LEVEL Log level (debug, info, warn, error)
export_format string json YOLO_AUDIT__EXPORT_FORMAT Export format (json, markdown)
retention_days int 30 YOLO_AUDIT__RETENTION_DAYS Days to retain logs
include_tokens bool true YOLO_AUDIT__INCLUDE_TOKENS Track token usage
include_costs bool true YOLO_AUDIT__INCLUDE_COSTS Track API costs 
audit:
  enabled: true
  log_level: info
  export_format: json
  retention_days: 30
  include_tokens: true
  include_costs: true

mcp

MCP server configuration.

Option Type Default Env Var Description
transport string stdio YOLO_MCP__TRANSPORT Transport type (stdio, http)
http.port int 8080 YOLO_MCP__HTTP__PORT HTTP port
http.host string 127.0.0.1 YOLO_MCP__HTTP__HOST HTTP host
mask_errors bool true YOLO_MCP__MASK_ERRORS Hide internal errors 
mcp:
  transport: stdio
  http:
    port: 8080
    host: "127.0.0.1"
  mask_errors: true

tools

External CLI tool integration configuration. YOLO can delegate tasks to external tools like Claude Code and Aider.

Option Type Default Env Var Description
claude_code.enabled bool false YOLO_TOOLS__CLAUDE_CODE__ENABLED Enable Claude Code integration
claude_code.path string None YOLO_TOOLS__CLAUDE_CODE__PATH Custom binary path
claude_code.timeout int 300 YOLO_TOOLS__CLAUDE_CODE__TIMEOUT Execution timeout (seconds)
claude_code.output_format string json YOLO_TOOLS__CLAUDE_CODE__OUTPUT_FORMAT Output format (json, text)
claude_code.extra_args list [] - Additional CLI arguments
aider.enabled bool false YOLO_TOOLS__AIDER__ENABLED Enable Aider integration
aider.path string None YOLO_TOOLS__AIDER__PATH Custom binary path
aider.timeout int 300 YOLO_TOOLS__AIDER__TIMEOUT Execution timeout (seconds)
aider.output_format string json YOLO_TOOLS__AIDER__OUTPUT_FORMAT Output format (json, text)
aider.extra_args list [] - Additional CLI arguments 
tools:
  claude_code:
    enabled: true
    path: null  # Use PATH lookup
    timeout: 300
    output_format: json
    extra_args: []
  aider:
    enabled: false
    path: null
    timeout: 300
    output_format: json
    extra_args: []

Environment Variables:

export YOLO_TOOLS__CLAUDE_CODE__ENABLED=true
export YOLO_TOOLS__CLAUDE_CODE__TIMEOUT=600
export YOLO_TOOLS__AIDER__ENABLED=true

Benefits of External Tool Integration

  • Security: Tools use their own authentication (no API key sharing)
  • Features: Access tool-specific capabilities (Claude Code plan mode, MCP client, web search)
  • Flexibility: Configure tools independently per project

Checking Tool Status 

yolo tools           # Show tool availability
yolo tools status    # Same as above
yolo tools --json    # JSON output for scripting

Environment Variables

All configuration can be overridden via environment variables using the YOLO_ prefix with __ as the nested delimiter.

Syntax 

YOLO_<SECTION>__<KEY>=<VALUE>

Examples 

# Project name
export YOLO_PROJECT_NAME=my-api

# LLM settings
export YOLO_LLM__PROVIDER=hybrid
export YOLO_LLM__CHEAP_MODEL=gpt-5.2-instant
export YOLO_LLM__PREMIUM_MODEL=claude-sonnet-4-20250514
export YOLO_LLM__BEST_MODEL=claude-opus-4-5-20251101
export YOLO_LLM__OPENAI__API_KEY=sk-proj-...
export YOLO_LLM__HYBRID__ENABLED=true

# Quality thresholds
export YOLO_QUALITY__TEST_COVERAGE_THRESHOLD=0.9
export YOLO_QUALITY__GATE_PASS_THRESHOLD=0.8

# Memory
export YOLO_MEMORY__PERSIST_PATH=/custom/path

# Agent settings
export YOLO_AGENTS__MAX_ITERATIONS=15
export YOLO_AGENTS__TIMEOUT_SECONDS=600

# Orchestrator
export YOLO_ORCHESTRATOR__HUMAN_ESCALATION_ENABLED=false

# MCP
export YOLO_MCP__TRANSPORT=http
export YOLO_MCP__HTTP__PORT=9000

# Brownfield scanning
export YOLO_BROWNFIELD__SCAN_DEPTH=4
export YOLO_BROWNFIELD__INTERACTIVE=false

# GitHub automation
export YOLO_GITHUB__TOKEN=ghp_...
export YOLO_GITHUB__REPOSITORY=bbengt1/yolo-developer

List Values

For list configuration (like blocking_gates), use comma-separated values:

export YOLO_QUALITY__BLOCKING_GATES=testability,architecture_validation,definition_of_done

Configuration Priority

When the same setting is defined in multiple places, the priority is:

  1. Environment variables (highest)
  2. yolo.yaml
  3. Built-in defaults (lowest)

Example 

# yolo.yaml
quality:
  test_coverage_threshold: 0.8

# Environment
export YOLO_QUALITY__TEST_COVERAGE_THRESHOLD=0.9

Result: test_coverage_threshold = 0.9 (environment wins)

Configuration Validation

Validate Configuration 

yolo config validate

Output:

Validating yolo.yaml...
  ✓ Schema valid
  ✓ API keys configured
  ✓ Memory directory writable
  ✓ Models available
  ✓ All settings valid

Configuration is valid.

Common Validation Errors

Invalid threshold:

Error: quality.test_coverage_threshold must be between 0.0 and 1.0
  Current value: 1.5
  Valid range: 0.0 - 1.0

Missing API key:

Error: No LLM API key configured
  Set one of:
    - YOLO_LLM__OPENAI__API_KEY
    - YOLO_LLM__ANTHROPIC_API_KEY

Invalid model:

Error: Unknown model: gpt-5-turbo
  Valid OpenAI models: gpt-5.2-pro, gpt-5.2-thinking, gpt-5.2-instant
  Valid Anthropic models: claude-3-opus-*, claude-3-sonnet-*, claude-3-haiku-*

Configuration Profiles

Development Profile 

# yolo.dev.yaml
project_name: my-api-dev

llm:
  cheap_model: gpt-5.2-instant  # Cheaper for dev
  premium_model: gpt-5.2-thinking
  best_model: gpt-5.2-pro

quality:
  test_coverage_threshold: 0.6  # Relaxed for iteration
  gate_pass_threshold: 0.5

agents:
  max_iterations: 5  # Faster iterations
  timeout_seconds: 120

audit:
  log_level: debug  # Verbose logging

Production Profile 

# yolo.prod.yaml
project_name: my-api-prod

llm:
  cheap_model: gpt-5.2-instant
  premium_model: claude-sonnet-4-20250514
  best_model: claude-opus-4-5-20251101

quality:
  test_coverage_threshold: 0.95  # Strict
  gate_pass_threshold: 0.9

agents:
  max_iterations: 15
  timeout_seconds: 600

audit:
  log_level: info
  retention_days: 90  # Longer retention

Using Profiles 

# Use specific config file
yolo --config yolo.dev.yaml run

# Or via environment
export YOLO_CONFIG_PATH=yolo.prod.yaml
yolo run

Complete Example 

# yolo.yaml - Complete configuration example

# Project identification
project_name: user-management-api

# LLM configuration
llm:
  provider: auto
  cheap_model: gpt-5.2-instant
  premium_model: claude-sonnet-4-20250514
  best_model: claude-opus-4-5-20251101
  # API keys via environment variables
  openai:
    cheap_model: gpt-5.2-instant
    premium_model: gpt-5.2-thinking
    code_model: gpt-5.2-pro
    reasoning_model: gpt-5.2-pro
  hybrid:
    enabled: false
    routing:
      code_generation: openai
      code_review: openai
      architecture: anthropic
      analysis: anthropic
      documentation: openai
      testing: openai

# Quality gates
quality:
  test_coverage_threshold: 0.85
  gate_pass_threshold: 0.75
  blocking_gates:
    - testability
    - ac_measurability
    - architecture_validation
    - definition_of_done
  warning_gates:
    - code_complexity

# Memory configuration
memory:
  persist_path: .yolo/memory
  vector_store_type: chromadb
  graph_store_type: json

# Agent configuration
agents:
  max_iterations: 10
  timeout_seconds: 300
  enable_caching: true
  cache_ttl_hours: 24
  parallel_stories: 1

  # Per-agent overrides
  dev:
    max_iterations: 15
    timeout_seconds: 600

  tea:
    timeout_seconds: 180

# Orchestration
orchestrator:
  max_agent_retries: 3
  circular_detection_threshold: 5
  human_escalation_enabled: true
  checkpoint_enabled: true
  checkpoint_interval: 5

# Audit and logging
audit:
  enabled: true
  log_level: info
  export_format: json
  retention_days: 30
  include_tokens: true
  include_costs: true

# MCP server
mcp:
  transport: stdio
  mask_errors: true

Next Steps

Table of contents

Back to top

Copyright © 2024-2026 YOLO Developer. Distributed under the MIT License.