Claude's Clock

Autonomous agent framework for executing development tasks using Ollama.

Setup

1. Ensure Ollama is running: ollama serve
2. Python 3.9+
3. Install: pip install ollama

Usage

from pathlib import Path
from claudeclockwork.agents.base_agent import BaseAgent
from claudeclockwork.agents.orchestrator import TaskOrchestrator

# Create orchestrator
orchestrator = TaskOrchestrator(Path.cwd())

# Define small task
task = {
    "id": "task-1",
    "description": "Generate unit tests",
    "agent_type": "test_generator",
    "files": ["src/module.py"],
}

# Execute
result = orchestrator.execute_task(task)
print(result)

Key Principles

1. Small Tasks Only: Each task must be atomic (1 function, 1 file, single concern)
2. Fast Models: Use gemma3:latest by default
3. Hybrid Execution: Python for structure, Ollama for content
4. Autonomous: No human decision gates
5. Self-Committing: Agents commit their own work

Agent Types

• test_generator: Generate unit tests
• doc_fixer: Fix documentation
• code_fixer: Fix bugs in code
• test_runner: Execute tests and report

Configuration

Edit claudeclockwork/config/ollama_config.py to:

• Change model selection
• Set Ollama host/port
• Configure timeout
• Enable/disable decomposition

Examples

See examples/ folder for working agent implementations.