Problem
Many Rust developers (and the broader local-AI community) prefer running models completely offline with Ollama. Right now RustifyMyClaw only supports the official Claude Code, Codex, and Gemini CLI tools. Users who have ollama installed and run models like llama3.2, qwen2.5-coder, phi4, deepseek-r1, etc. cannot use RustifyMyClaw to talk to them
Proposed solution
Add a new backend ollama that:
- Uses the existing CliBackend trait (see
src/backend/ and How to Add a New Backend)
- Supports Interactive chat mode (
ollama run <model>)
- simplest, works great for normal conversation.
Alternatives considered
Using a Rust Ollama client crate (ollama-rs, ollama-client, etc.) directly instead of spawning the CLI — cleaner long-term but adds a dependency and deviates from the current "proxy to existing CLI" architecture.
Problem
Many Rust developers (and the broader local-AI community) prefer running models completely offline with Ollama. Right now RustifyMyClaw only supports the official Claude Code, Codex, and Gemini CLI tools. Users who have ollama installed and run models like llama3.2, qwen2.5-coder, phi4, deepseek-r1, etc. cannot use RustifyMyClaw to talk to them
Proposed solution
Add a new backend
ollamathat:src/backend/and How to Add a New Backend)ollama run <model>)Alternatives considered
Using a Rust Ollama client crate (ollama-rs, ollama-client, etc.) directly instead of spawning the CLI — cleaner long-term but adds a dependency and deviates from the current "proxy to existing CLI" architecture.