Story Forge is a digital platform that transforms books into interactive text-adventure
games using local large language models (LLMs). This iteration adapts Mutiny Aboard the
HMS Bounty, a historical account of a late 18th-century revolt aboard a British merchant
vessel, allowing players to engage with the story dynamically. The platform employs a "Story
Card" system, a structured summary that the model references at each turn to maintain
narrative coherence. This approach ensures that character actions, plot progression, and story
structure remain aligned with the original material while still allowing for player agency and
variation in how events unfold.

Through iterative prototyping and testing, Story Forge demonstrates both the
strengths and limitations of LLMs in narrative design. The model enables dynamic storytelling
by allowing players to engage with the narrative in multiple ways, supporting new plot
branches and alternative story arcs. However, maintaining strict adherence to the source
material remains a challenge, as the AI occasionally deviates from the structured plot despite
the Story Card constraints. While the model effectively processes player input and generates
cohesive responses, its output (currently) lacks the nuance of human-authored storytelling,
sometimes resulting in dialogue and descriptions that feel repetitive or lacking in depth.

These findings highlight the potential of AI-driven narrative systems to create
interactive adaptations of existing stories while also revealing the complexities of balancing
structure and flexibility in procedural storytelling. This research provides a foundation for
further exploration into AI-assisted narrative design and offers insight into how large language
models can be used to enhance engagement, personalization, and interactivity in digital
storytelling.

• https://www.matthewnazarian.com/projects...
• https://github.com/mattocad/Story-Forge-...
• https://github.com/mattocad/Story-Forge-...