config.jsonc Reference

All options, default values, and usage examples for the OMC configuration file (config.jsonc)

Configuration File Location

Scope	Path
User (global)	`~/.config/claude-omc/config.jsonc`
Project	`.claude/omc.jsonc`

Settings are merged in the order: defaults → user config → project config → environment variables. Later settings override earlier ones.

agents

Specifies the model for each agent.

{
  "agents": {
    "omc": { "model": "opus" },        // Main orchestrator
    "explore": { "model": "haiku" },    // Codebase exploration
    "analyst": { "model": "opus" },     // Requirements analysis
    "planner": { "model": "opus" },     // Task planning
    "architect": { "model": "opus" },   // System design
    "debugger": { "model": "sonnet" },  // Debugging
    "executor": { "model": "sonnet" },  // Code implementation
    "verifier": { "model": "sonnet" },  // Verification
    "securityReviewer": { "model": "sonnet" }, // Security review
    "codeReviewer": { "model": "opus" },       // Code review
    "testEngineer": { "model": "sonnet" },     // Test strategy
    "designer": { "model": "sonnet" },  // UI/UX design
    "writer": { "model": "haiku" },     // Documentation
    "qaTester": { "model": "sonnet" },  // QA testing
    "scientist": { "model": "sonnet" }, // Data analysis
    "tracer": { "model": "sonnet" },    // Causal tracing
    "gitMaster": { "model": "sonnet" }, // Git operations
    "codeSimplifier": { "model": "opus" },     // Code simplification
    "critic": { "model": "opus" },      // Plan review
    "documentSpecialist": { "model": "sonnet" } // Document lookup
  }
}

Agent names with hyphens (-) are converted to camelCase in the config. Example: security-reviewer → securityReviewer, code-reviewer → codeReviewer

features

Feature toggles.

{
  "features": {
    "parallelExecution": true,        // Parallel agent execution
    "lspTools": true,                 // LSP tool integration
    "astTools": true,                 // AST tools (ast-grep)
    "continuationEnforcement": true,  // Force task completion
    "autoContextInjection": true      // Auto context injection
  }
}

Option	Default	Description
`parallelExecution`	`true`	Run multiple agents simultaneously
`lspTools`	`true`	Enable Language Server Protocol tools
`astTools`	`true`	Enable ast-grep based AST tools
`continuationEnforcement`	`true`	Continue if there are incomplete tasks
`autoContextInjection`	`true`	Auto-inject CLAUDE.md/AGENTS.md

magicKeywords

Allows customizing magic keyword triggers.

{
  "magicKeywords": {
    "ultrawork": ["ultrawork", "ulw", "uw"],
    "search": ["search", "find", "locate"],
    "analyze": ["analyze", "investigate", "examine"],
    "ultrathink": ["ultrathink", "think", "reason", "ponder"]
  }
}

Specify a string array for each category. When user input contains one of these strings, the corresponding mode is activated.

Only the 4 categories above can be changed via magicKeywords. Keywords like autopilot, ralph, and ccg are hardcoded in the keyword-detector hook and cannot be changed via config.

routing

Model routing configuration.

{
  "routing": {
    "enabled": true,           // Enable intelligent model routing
    "defaultTier": "MEDIUM",   // Default tier (LOW | MEDIUM | HIGH)
    "forceInherit": false,     // Force parent model inheritance
    "escalationEnabled": true, // Allow escalation
    "maxEscalations": 2        // Maximum number of escalations
  }
}

Option	Default	Description
`enabled`	`true`	Whether to enable model routing
`defaultTier`	`"MEDIUM"`	Default tier applied when no rules match
`forceInherit`	`false`	When `true`, all agents inherit the parent model
`escalationEnabled`	`true`	Allow automatic upgrades based on task complexity
`maxEscalations`	`2`	Maximum number of escalations

forceInherit is automatically set to true when using CC Switch, custom ANTHROPIC_BASE_URL, AWS Bedrock, or Google Vertex AI. This is because passing Claude-specific model names (sonnet, opus, haiku) to these platforms causes errors.

permissions

Permission configuration.

{
  "permissions": {
    "allowBash": true,          // Allow Bash command execution
    "allowEdit": true,          // Allow file editing
    "allowWrite": true,         // Allow file creation
    "maxBackgroundTasks": 5     // Maximum background tasks (1~50)
  }
}

mcpServers

MCP server configuration.

{
  "mcpServers": {
    "exa": {
      "enabled": true,
      "apiKey": "your-exa-api-key"  // Or use EXA_API_KEY env var
    },
    "context7": {
      "enabled": true
    }
  }
}

externalModels

External model provider configuration.

{
  "externalModels": {
    "defaults": {
      "codexModel": "gpt-5.3-codex",          // Default Codex model
      "geminiModel": "gemini-3.1-pro-preview"  // Default Gemini model
    },
    "fallbackPolicy": {
      "onModelFailure": "provider_chain",  // provider_chain | cross_provider | claude_only
      "allowCrossProvider": false,
      "crossProviderOrder": ["codex", "gemini"]
    }
  }
}

delegationRouting

Delegation routing configuration to external models (opt-in feature).

{
  "delegationRouting": {
    "enabled": false,              // Disabled by default
    "defaultProvider": "claude",   // claude | codex | gemini
    "roles": {
      "executor": {
        "provider": "codex",
        "tool": "Task"
      }
    }
  }
}

startupCodebaseMap

Codebase map generation settings at session start.

{
  "startupCodebaseMap": {
    "enabled": true,
    "maxFiles": 200,
    "maxDepth": 4
  }
}

taskSizeDetection

Automatic task size detection settings.

{
  "taskSizeDetection": {
    "enabled": true,
    "smallWordLimit": 50,                    // Tasks with fewer words are small tasks
    "largeWordLimit": 200,                   // Tasks with more words are large tasks
    "suppressHeavyModesForSmallTasks": true  // Suppress heavy modes for small tasks
  }
}

Full Default Configuration

Below are all of OMC's default configuration values.

{
  "agents": {
    "omc": { "model": "opus" },
    "explore": { "model": "haiku" },
    "analyst": { "model": "opus" },
    "planner": { "model": "opus" },
    "architect": { "model": "opus" },
    "debugger": { "model": "sonnet" },
    "executor": { "model": "sonnet" },
    "verifier": { "model": "sonnet" },
    "securityReviewer": { "model": "sonnet" },
    "codeReviewer": { "model": "opus" },
    "testEngineer": { "model": "sonnet" },
    "designer": { "model": "sonnet" },
    "writer": { "model": "haiku" },
    "qaTester": { "model": "sonnet" },
    "scientist": { "model": "sonnet" },
    "tracer": { "model": "sonnet" },
    "gitMaster": { "model": "sonnet" },
    "codeSimplifier": { "model": "opus" },
    "critic": { "model": "opus" },
    "documentSpecialist": { "model": "sonnet" }
  },
  "features": {
    "parallelExecution": true,
    "lspTools": true,
    "astTools": true,
    "continuationEnforcement": true,
    "autoContextInjection": true
  },
  "magicKeywords": {
    "ultrawork": ["ultrawork", "ulw", "uw"],
    "search": ["search", "find", "locate"],
    "analyze": ["analyze", "investigate", "examine"],
    "ultrathink": ["ultrathink", "think", "reason", "ponder"]
  },
  "routing": {
    "enabled": true,
    "defaultTier": "MEDIUM",
    "forceInherit": false,
    "escalationEnabled": true,
    "maxEscalations": 2
  },
  "permissions": {
    "allowBash": true,
    "allowEdit": true,
    "allowWrite": true,
    "maxBackgroundTasks": 5
  },
  "externalModels": {
    "defaults": {
      "codexModel": "gpt-5.3-codex",
      "geminiModel": "gemini-3.1-pro-preview"
    },
    "fallbackPolicy": {
      "onModelFailure": "provider_chain",
      "allowCrossProvider": false,
      "crossProviderOrder": ["codex", "gemini"]
    }
  },
  "delegationRouting": {
    "enabled": false,
    "defaultProvider": "claude",
    "roles": {}
  },
  "startupCodebaseMap": {
    "enabled": true,
    "maxFiles": 200,
    "maxDepth": 4
  },
  "taskSizeDetection": {
    "enabled": true,
    "smallWordLimit": 50,
    "largeWordLimit": 200,
    "suppressHeavyModesForSmallTasks": true
  }
}

config.jsonc Reference

On this page