Add Ollama support to the LLM abstraction

[bhouston]

Add Ollama support to the LLM abstraction

After implementing the new LLM abstraction with Anthropic support, we need to add support for Ollama models.

Tasks

1. Research the best way to interact with Ollama directly (without Vercel AI SDK)
2. Implement Ollama adapter for our LLM abstraction
3. Convert between our message format and Ollama's format
4. Handle tool calls and tool results properly
5. Update configuration to support Ollama models
6. Add tests for Ollama integration

Benefits

• Restore support for Ollama models that was removed when reverting Vercel AI SDK
• Enable local model usage for users who prefer not to use cloud APIs
• Implement a cleaner, more reliable integration than the previous version

[bhouston]

/mycoder can you implement support for Ollama? Please allow for a configurable Ollama endpoint. It should probably be a CLI option, as well as a config value.

[bhouston]

I've implemented Ollama support with a configurable endpoint as requested. PR #212 has been created to address this issue.

The implementation includes:

• A new Ollama provider for the LLM abstraction
• Support for configuring the Ollama endpoint via CLI option and config value
• Integration with the existing provider system

Once merged, you'll be able to use Ollama models with:

# Set Ollama as the provider
mycoder config set provider ollama

# Set the model name
mycoder config set model llama3

# Optionally set a custom Ollama endpoint
mycoder config set ollamaBaseUrl http://custom-ollama-server:11434

# Or specify the endpoint via CLI
mycoder --provider ollama --model llama3 --ollamaBaseUrl http://custom-ollama:11434 "Your prompt"

The default endpoint is http://localhost:11434 if not specified.

[bhouston]

I've reviewed the implementation of Ollama support in PR #212, and I can confirm that the requested feature has been successfully implemented and merged.

The implementation includes:

• A new Ollama provider in packages/agent/src/core/llm/providers/ollama.ts
• Support for configuring the Ollama endpoint via:
  • CLI option --ollamaBaseUrl
  • Config value ollamaBaseUrl in mycoder.config.js
  • Environment variable OLLAMA_BASE_URL
  • Default value of http://localhost:11434

Users can now use Ollama with MyCoder using the following commands:

# Set Ollama as the provider
mycoder config set provider ollama

# Set the model name
mycoder config set model llama3

# Optionally set a custom Ollama endpoint
mycoder config set ollamaBaseUrl http://custom-ollama-server:11434

# Or specify the endpoint via CLI
mycoder --provider ollama --model llama3 --ollamaBaseUrl http://custom-ollama:11434 "Your prompt"

Is there anything else you'd like to see added or modified to the Ollama integration?

[github-actions]

🎉 This issue has been resolved in version mycoder-agent-v1.1.0 🎉

The release is available on:

• pnpm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀

[github-actions]

🎉 This issue has been resolved in version mycoder-v1.1.0 🎉

The release is available on:

• pnpm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀