Currently, when performing a capture list action, the list and its fields are labeled generically (e.g., “List 1”, “Label 1”, “Label 2”), and users must rename them manually. We need to automatically generate meaningful list and field names based on context using an LLM. It should support multiple providers (Anthropic, OpenAI, OpenRouter, Ollama, etc.) and fall back to the default labeling if the LLM call fails.

Currently, when performing a capture list action, the list and its fields are labeled generically (e.g., “List 1”, “Label 1”, “Label 2”), and users must rename them manually. We need to automatically generate meaningful list and field names based on context using an LLM. It should support multiple providers (Anthropic, OpenAI, OpenRouter, Ollama, etc.) and fall back to the default labeling if the LLM call fails.