You connect an AI assistant to your Dramatica account. It can find your Story, retrieve the Storyform, look up the Main Character Problem, and pull examples from stories that share the same structural choice.
So why would you ever leave that conversation and return to Narrova?
It’s a fair question, especially now that assistants such as Claude and ChatGPT can reach beyond their own training and work with the tools you authorize. Once they can see your story, the distance between a general frontier model and a dedicated story-development system can appear very small.
The distance becomes visible the moment you ask a harder question: What if the Storyform doesn’t exist yet?
MCP gives an agent a bounded way to retrieve, validate, and update parts of your Dramatica work. Narrative Intelligence coordinates the deeper process of discovering, testing, and preserving what the story can actually support.
A Conversation Has to Leave Something Behind
Story development rarely begins with a clean set of structural choices. It begins with fragments: a character who won’t let go, an ending that feels right for reasons the writer can’t yet explain, a relationship that matters more than the plot seems willing to admit. The writer talks, tests, rejects, circles back, and slowly recognizes what the story is about.
Narrova is built for that movement. You can pour out a stream of ideas in ordinary language while it keeps the active Story and Storyform in view. Its Story Development Ledger becomes the running notebook of the work: what appears settled, what remains provisional, which questions are open, and which decisions still belong to the writer.
That notebook is useful because a transcript is not the same thing as a story. The crucial decision may be hiding inside forty minutes of exploration, surrounded by alternatives you already abandoned. The Ledger gives the developing story a current state without pretending every confident sentence became canon.
Then there are the moments when you really do know.
You know the Story Goal. You know the Main Character Problem. You know what the Relationship Story is about, or what kind of conflict belongs in one particular Storypoint. At that point, you can ask Narrova to save the choice as a discrete component attached to the Storyform.
The conversation has produced something durable. You have stuck a pin in the story.
Some Writers Think in Conversation. Others Need to See the Shape.
This is where Subtxt enters the workflow.
Subtxt is the structured application we began building in 2017. Where Narrova gives you room to think through a story conversationally, Subtxt lets you see the concrete pieces of the Storyform: the Four Throughlines, the Storypoints attached to them, the Storybeats that bring those points into time, and the Storytelling you have developed around each choice.
“The two modules work very well together, and you can always bounce back-and-forth between Subtxt and Narrova as you develop.”
— Jim Hull, Dramatica Community forum, September 2026
That back-and-forth matters because writers do not all recognize certainty in the same way. One writer needs the freedom of a long conversation before the real idea appears. Another needs to look at the whole structure and see which boxes are filled, which are empty, and which no longer agree with one another.
There is no virtue in choosing one style over the other. Narrova and Subtxt work against the same developing Storyform, so the writer can move between exploration and inspection. A choice discovered in conversation can become a visible component. A gap noticed in Subtxt can become the next question brought back to Narrova.
The story stays coherent while the interface changes.
MCP Lets the Story Travel
Model Context Protocol, or MCP, solves a different problem. It gives an AI assistant permission to use tools and retrieve information from another system. In this case, Dramatica MCP lets a connected agent reach selected parts of the Dramatica Narrative Platform instead of answering only from the general patterns contained in the model itself.
That is a meaningful capability. A writer can ask Claude or ChatGPT to list the Stories in their account, retrieve a Storyform, inspect a saved Narrative Aspect, research canonical examples, or look up a precise point of Dramatica theory. The MCP can also validate a set of proposed Storyform choices and, with explicit confirmation, save supported changes.
“The MCP gives your frontier agents access to many of Dramatica’s tools but definitely not all of them.”
— Jim Hull, Dramatica Community forum, September 2026
The boundary in that sentence is the point. MCP gives the outside agent a carefully permissioned way to work with Dramatica material. It does not move the whole of Narrova’s Narrative Intelligence into the outside conversation.
This makes MCP especially useful once the writer has something established. Perhaps the Storyform is settled and you want to explore a production question in Claude. Perhaps you want ChatGPT to compare your Main Character Problem with canonical examples. Perhaps another agent has a research or planning tool you like, and you want it to begin from the real Storyform instead of a summary pasted from memory.
MCP lets the story travel without forcing it to become a loose paragraph of context every time it crosses into a new tool.
A Screenplay Reveals the Boundary
Imagine that you have just finished a screenplay draft and want story coverage.
If you already possess a trustworthy Storyform, MCP can make that structure available to an outside agent. The agent can discuss the draft in light of the Story Goal, the Main Character Problem, the Four Throughlines, and the other choices you have saved. It can help you iterate, research, or examine possibilities from within its own strengths.
But suppose you do not know the Storyform. You have a screenplay and a suspicion that something is wrong.
Now the work is no longer retrieval. Someone has to read the draft through four distinct viewpoints of conflict: what everyone is dealing with, what the central character experiences personally, the alternative Perspective challenging that character, and the relationship changing between them. Dramatica calls these the Objective Story, Main Character, Influence Character, and Relationship Story Throughlines.
Those readings have to be compared. A plausible Domain arrangement may fail when the Concerns are added. A compelling interpretation of the Main Character may require an Objective Story the screenplay never supplies. The analyst has to propose candidates, test them against Dramatica’s Story Engine, reject combinations that cannot hold together, return to the evidence, and try again.
Narrova’s Story Coverage is built to read that evidence, compare likely candidate structures, and identify revision priorities. When the writer needs one most defensible candidate, the separate Find Best Storyform workflow carries the analysis to that result. The current MCP gives an outside agent useful Dramatica tools, but it does not reproduce either complete upload-led process.
The distinction is easy to miss because a frontier model can still produce coverage. It has absorbed enormous amounts of Storytelling: scripts, novels, criticism, craft advice, summaries, and the language people use when talking about stories. It can recognize patterns and offer intelligent notes.
Storyforming asks a different kind of question. It asks what conceptual arrangement of conflict makes all four viewpoints part of one argument. Fluent Storytelling knowledge can suggest an answer; it cannot, by itself, establish that the answer belongs to a coherent Dramatica Storyform.
For a closer look at that analysis loop, see A Storyform Gives Story Coverage Somewhere to Land.
The Writer Still Chooses the Distance
Once Narrova has helped you find the strongest Storyform candidate, the possibilities widen again. You can keep developing inside Narrova, open the components in Subtxt, or carry the established context through MCP into Claude, ChatGPT, Codex, or another agent.
Moving outward may be exactly what the work needs. Another system may offer a tool you prefer, a different conversational feel, or a useful way to research and organize the next stage. You may want to step away from Dramatica terminology for a while and let the story breathe in another creative environment.
You should know what changes when you do. The outside agent may have access to Dramatica facts and your saved story context, but the conversation is no longer taking place inside Narrova’s complete theory-grounded development process. That is not a failure of MCP. It is the freedom MCP is meant to provide.
The useful question is not which system gets to own the story. It is which kind of help the writer needs right now.
Use Narrova when you need an objective partner to discover, test, and preserve what the story means. Use Subtxt when you want to see those decisions in concrete form and work through the Storyform directly. Use MCP when you want established Dramatica context and selected tools to accompany you into another agent.
That is the freedom of the larger system: the writer can change rooms without pretending each room does the same work. The conversation can wander, the Storyform can hold, and the draft can return to the page with its meaning intact.
Continue with Connect Dramatica to an AI Assistant (MCP), Story Development, Introduction to Subtxt, and AI Shouldn’t Write the Story for You. It Can Still Help You Find It..