Services

ChatService implementations handle the tool-call loop and communication with LLM providers.

Navigation

• ChatService: abstract base class with tool-call loop, prompt loading
• OpenAI: OpenAI-compatible streaming implementation
• Inline reasoning: opt-in parsing of explicit reasoning markers in content
• Inline tool calls: opt-in parsing of explicit <tool_call> blocks in content
• Interrupt: abort in-flight requests and re-send
• Custom provider: adding a new provider