Show HN: ThoughtDAG – An editable context graph for LLM conversations

What about cache? When you change the context the prefill stage will be much slower?

[flagged]