Skip to content

configuration.md

Latest commit

 

History

History
95 lines (81 loc) · 2.09 KB

configuration.md

File metadata and controls

95 lines (81 loc) · 2.09 KB

Configuration

File Locations

~/.config/karl/
├── karl.json        # Main config
├── tools/           # Custom tools
├── skills/          # Skill definitions
└── hooks/           # Hook scripts

Project-level .karl.json overrides global config.

Config Schema

{
  "defaultModel": "fast",
  "models": {
    "fast": {
      "provider": "openrouter",
      "model": "anthropic/claude-sonnet-4-20250514"
    },
    "smart": {
      "provider": "anthropic",
      "model": "claude-sonnet-4-20250514"
    }
  },
  "providers": {
    "openrouter": {
      "type": "openai",
      "baseUrl": "https://openrouter.ai/api/v1",
      "apiKey": "${OPENROUTER_API_KEY}"
    },
    "anthropic": {
      "type": "anthropic",
      "apiKey": "${ANTHROPIC_API_KEY}"
    }
  },
  "tools": {
    "enabled": ["bash", "read", "write", "edit"],
    "custom": ["~/.config/karl/tools/*.ts"]
  },
  "volley": {
    "maxConcurrent": 3,
    "retryAttempts": 3,
    "retryBackoff": "exponential"
  }
}

Context Loading

Automatic project context from:

  • CLAUDE.md, AGENTS.md, COPILOT.md
  • .cursorrules, .github/copilot-instructions.md
  • .karl/context.md (project-specific)

Explicit context:

karl run --context "You are a security expert" "review this"
karl run --context-file system-prompt.txt "analyze the architecture"

No session state. Each invocation is fresh. For multi-turn work, use a full coding agent and delegate to Karl for side tasks.

Hooks

Run custom logic at key points:

Hook When
pre-task Before each task starts
post-task After each task completes
pre-tool Before tool execution
post-tool After tool execution
on-error When a task fails 
// ~/.config/karl/hooks/audit.ts
export default {
  'post-tool': async (event) => {
    if (event.tool === 'bash') {
      await appendToLog(`[${new Date()}] ${event.input.command}`);
    }
  }
};

Use cases:

  • Audit logging for compliance
  • Notifications (Slack, email) on completion
  • Metrics/telemetry collection
  • Custom retry logic