Games treat dialogue as scripting. Film and TV treat narrative as production, with scripts, cast, recording sessions, reviews, and certification. StoryBonsai brings that model to games.

Media duration tracking

Automatic duration calculation from linked media assets. Know your recording schedule before you book a studio. Know your running time before you sign a contract.

Test suite for narrative

Expression tests pin condition behavior. Define cases (these variables, this play history, this inventory, this location, this time of day) and lock in the expected outcome. The editor suggests boundary scenarios from the variables each condition reads, so coverage doesn't depend on someone remembering the edge cases. Catches "fixed once, broken later" regressions before they ship.

The whole suite runs in seconds. Failing tests highlight inline on the entity that owns them. Tests live with the field they describe, so renaming the field doesn't break the test.

Thousands of playthroughs before QA plays one

Play thousands of versions of your story, reproducible by seed. Four strategies, each exploring the story differently:

Random. Uniform exploration.
Defaults. Always the first-authored choice; verifies the golden path.
Coverage-guided. Pushes into under-explored regions.
Greedy goal-seeking. Drives toward a target scene or variable state.

The tester walks the world alongside choices: moving between locations, starting encounters, advancing time, and completing activities. It catches what a static scan cannot: unreachable nodes, variable-bounds violations, duration mismatches, infinite loops. Coverage heatmaps show where the narrative is under-tested.

What does a path-test report include?

Every path-test run produces a report covering everything observed across the playthroughs: issues with traces, coverage, a skill-check breakdown (roll counts, pass rate, DC distribution), a dead-end analysis grouped by node with variable-state history, location and encounter reachability, and a scene call graph showing which scenes lead where and how often.

Reports export as markdown for human reviewers or structured JSON for dashboards and build archives, and every run writes to a reports/ folder alongside the project: the latest results, an accumulating run log, and a readable summary. Hand it to your publisher, archive it with the build, review it over coffee.

Project Validator

Every broken reference, unreachable branch, out-of-bounds variable, and expression that returns the wrong type for its target variable surfaces in one panel. Production catches issues before QA does.

How does the translation pipeline work?

Every project carries a source locale and any number of target locales as overlays. The Localization view tracks coverage per string, flags entries that went stale when the source changed, and carries needs-review markers through the whole round trip. Handoffs export in CSV, JSON, or XLIFF, filtered by status (all, stale only, untranslated only) and scope (project or single scene). Imports merge through a diff that separates clean updates, stale rows with per-item decisions, and orphaned keys surfaced instead of silently dropped.

Structural edits are locked to the source locale, so a translation pass can't change the story. Point the path tester at a target locale and every automated run doubles as translation QA, reporting untranslated strings alongside the other issues.

Story beats for recap and analytics

Writers tag narrative moments in video and audio. The engine remembers which beats a player has seen and at what importance. Build "previously on…" recaps, character-aware ambience, or narrative analytics from data already in your story.

Git-native collaboration

One file per node, per scene, per location, per item, per skill. Small, focused diffs. Zero merge conflicts between writers editing different entities. Locale overlays follow the same layout, and your team's git workflow stays as it is.

Platform and engine coverage

Engines: Unity, Godot, MonoGame via .NET Standard 2.1
Tested today: PC, Mac, Steam Deck
Coming: mobile and console

Single-player or multiplayer. Built into the runtime.

Ship multiplayer narrative without re-architecting. Split-screen or online, with state tracked per player.

Coming soon

In active development.

Embeddable web preview Share any conversation with anyone: producers, voice directors, stakeholders, QA. Just a URL, no install required. Click-through review for teams that can't run the editor.
Voice production pipeline CSV and markdown script export, recording session handoff, duration-source tracking on every line.
Content sensitivity tagging Map content warnings to nodes. Generate a summary for ratings-body submissions (ESRB, PEGI, BBFC).

Request Access →