ThoughtDAG lets users edit LLM context as a graph

Developer Chen Xia Chan built ThoughtDAG, an open-source prototype that represents LLM conversations as a directed acyclic graph. Users can branch, merge, and prune context by manipulating edges, which determine which nodes are included in the model's next request. The tool is local-first, stores data in IndexedDB, and supports Ollama and OpenAI-compatible endpoints.