< Portfolio

ChainForge

ChainForge is an open-source visual programming environment for battle-testing prompts to large language models (LLMs). With ChainForge, you can compare across models, prompts, and prompt parameters using built-in visualizations. Read more on the public GitHub page.

An ChainForge data flow diagram setup to evaluate prompt injection attacks.

I am developing ChainForge "in the open", as part of a critique of an academic atmosphere of closed-source or it-worked-for-the-user-study prototypes, where broader impacts of new systems are claimed but rarely actually achieved for the benefit of the wider public.

ChainForge is being developed with colleagues at Harvard CS: Elena Glassman, Priyan Vaithilingam, and Martin Wattenberg.