Skip to content

AI Coding Tools

Configure AI-powered coding assistants to use local LLM backends.

Overview

AI coding tools can connect to:

  • Local Ollama - Self-hosted, private
  • Local llama.cpp - Custom configurations
  • OpenAI-compatible APIs - Any compatible endpoint
  • Cloud APIs - Anthropic, OpenAI (fallback)

Tool Comparison

Tool Interface Local Model Support Best For
Claude Code CLI Via API proxy Anthropic CLI users
Aider CLI Native Ollama Git-integrated coding
Cline VS Code Native VS Code users
Continue.dev Multi-editor Native IDE integration

Feature Matrix

┌─────────────────────────────────────────────────────────────────┐
│                    AI Coding Tools                               │
├─────────────────────────────────────────────────────────────────┤
│                                                                  │
│  CLI Tools:                                                      │
│  ┌─────────────────────┐    ┌─────────────────────┐             │
│  │    Claude Code      │    │       Aider         │             │
│  ├─────────────────────┤    ├─────────────────────┤             │
│  │ ✓ Official Claude   │    │ ✓ Git-aware         │             │
│  │ ✓ Multi-file edits  │    │ ✓ Local models      │             │
│  │ ✓ Code execution    │    │ ✓ Auto-commits      │             │
│  │ ✓ File operations   │    │ ✓ Multiple LLMs     │             │
│  └─────────────────────┘    └─────────────────────┘             │
│                                                                  │
│  IDE Extensions:                                                 │
│  ┌─────────────────────┐    ┌─────────────────────┐             │
│  │       Cline         │    │    Continue.dev     │             │
│  ├─────────────────────┤    ├─────────────────────┤             │
│  │ ✓ VS Code           │    │ ✓ VS Code + JetBrains│            │
│  │ ✓ Plan/Act modes    │    │ ✓ Autocomplete      │             │
│  │ ✓ Local providers   │    │ ✓ Chat + Edit       │             │
│  │ ✓ MCP support       │    │ ✓ Local-first       │             │
│  └─────────────────────┘    └─────────────────────┘             │
│                                                                  │
└─────────────────────────────────────────────────────────────────┘

Development Workflow

                    ┌─────────────────┐
                    │  Code Question  │
                    └────────┬────────┘
                             │
            ┌────────────────┼────────────────┐
            │                │                │
            ▼                ▼                ▼
     ┌──────────┐     ┌──────────┐     ┌──────────┐
     │ Continue │     │   Aider  │     │   Cline  │
     │ (IDE)    │     │  (CLI)   │     │ (VS Code)│
     └────┬─────┘     └────┬─────┘     └────┬─────┘
          │                │                │
          └────────────────┴────────────────┘
                           │
                    ┌──────┴──────┐
                    │   Ollama    │
                    │ (Local LLM) │
                    └─────────────┘

Model Recommendations

Task Model Notes
Code completion DeepSeek Coder V2 16B Fast, accurate
Code chat Llama 3.3 70B Good reasoning
Refactoring Qwen 2.5 Coder 32B Balanced
Quick answers Mistral 7B Very fast

Quick Configuration

Environment Variables

# For OpenAI-compatible tools
export OPENAI_API_BASE=http://localhost:11434/v1
export OPENAI_API_KEY=not-needed

# For Ollama-native tools
export OLLAMA_HOST=http://localhost:11434

Ollama Setup

Ensure Ollama is running with your preferred model:

# Start Ollama
ollama serve

# Pull coding model
ollama pull deepseek-coder-v2:16b

# Verify
curl http://localhost:11434/v1/models

Topics

Model Configuration Examples

For Aider

# Using Ollama
aider --model ollama/deepseek-coder-v2:16b

# Using OpenAI-compatible
aider --openai-api-base http://localhost:8080/v1 --model llama3.3

For Continue.dev

{
  "models": [
    {
      "title": "DeepSeek Coder",
      "provider": "ollama",
      "model": "deepseek-coder-v2:16b"
    }
  ]
}

For Cline

Settings → Cline → API Provider → Ollama

See Also