[Question] Option to have add_episode without LLM extraction

[nivida]

I defined many custom entities and edges for relational data of N sources I want to ingest into Graphiti. The LLM extraction at the end is an overkill (slow and costly) for this case, and decreases the data quality because it gives me no deterministic proof that everything is linked correctly. I was able to get around it by further extending the custom extraction instructions. But I am still scared that the AI has a bad moment and doesn't do its job as defined in the prompt (we all had this in the past when working with LLMs).

A few questions:

• Are there plans to support the possibility to add_episode and add_episode_bulk to pass perfectly linked data, including cross-source data links?
• I also needed to implement a PendingLinks queue system to prevent cross-source webhook ingestion race conditions, in case, for example, the Task Tool emits the task event webhook after the Messenger message event webhook, where the task already been referenced. Are there plans to provide an out-of-the-box solution from this package?
• The Graphiti extraction pipeline ignores code snippets by default. If we, for example, want to ingest a specific procedure with code samples, this is quite a problem. Would be great to have an option to enable the inclusion of code snippets. We would be able to get around it if you were able to exclude the LLM extraction on the ingestion side. What are your thoughts?

I know many questions in one issue. But I hope I get some good answers so that we can, together, improve the Graphiti open source project.