ThoughtDAG

ThoughtDAG is an open-source tool for having AI conversations on an infinite canvas where each exchange becomes a node. Users can edit the context sent to language models, explore side questions in branches, and selectively bring past discussions into new queries. The interface visualizes conversation history as a graph rather than a linear thread, letting users see exactly what context will be sent to the model before each request.

Designed for researchers, writers, and anyone doing complex thinking with AI, ThoughtDAG addresses context pollution by letting users remove unwanted discussion paths while keeping earlier work intact. Version 0.5 adds a recall feature to find and reuse relevant past conversations, with optional integration of a decision model for ranking excerpts. The project indicates it was built with Claude Code.