Trace story paths
Find broken destinations, unreachable scenes, and a path that reaches a scene without its declared prerequisites.
Inspect branches, trace missing prerequisites, and prepare context-rich narrative reviews for Claude. Built for indie creators working with branching dialogue.
arrival → open_gate
The key was never acquired.
open_gate → ending_typo
No scene with this ID exists.
Small script edits can leave a broken branch or an item used before it is acquired. Inspect these structural problems, then prepare the full story context for a human-reviewed Claude analysis.
Find broken destinations, unreachable scenes, and a path that reaches a scene without its declared prerequisites.
Export world notes, dialogue, and branch findings in one review brief. Ask Claude for contradictions with scene IDs and source excerpts. You send the brief yourself.
Everything in this demo runs in your browser. Script text is not transmitted to a server or sent to an AI model.
Narrative QA · working prototype
Try the original four-scene example with three planted findings. Load the corrected version to see the difference. These checks use declared rules, not AI interpretation.
entry names the starting scene. choices links scenes with to. requires lists facts expected on scene entry; sets adds facts after its dialogue. All choices are unconditional. Facts persist: removal, numeric conditions, and timed events are not modeled. Blank choices marks an ending. JSON examples are editable. At most 200 scenes and 1,000 distinct path states are checked; incomplete analysis is explicitly flagged.
Download and inspect this brief before sending it to Claude yourself. It asks for continuity and character-knowledge findings with source excerpts. Semantic review and automatic Claude API integration are not performed by this website.
Try it now
Paste your own lines or load the example. Group them, inspect the result, and export it — all without leaving this page.
From preparation to translation
Add a scene and glossary, then copy the brief into your own Claude conversation. No API key or paid integration required by Lineweave.
1. Build and inspect the brief. 2. Paste it into Claude yourself. 3. Review meaning, character voice, and placeholders before using any translation. Input is cleared when this page reloads.
After Claude responds
Paste Claude’s JSON response to check line IDs, missing placeholders, and glossary matches against your current script. These are format checks; meaning and tone still need human review.
The example is hand-written with two intentional errors, not a Claude result. Loading it replaces the demo script and glossary. Glossary checks use exact text matching and may need your judgment.
About the project
Lineweave helps indie game creators review branching scripts before release. The current prototype checks declared story structure locally and prepares a Claude review brief. Localization preparation and format checks are included as supporting tools. You decide which findings need a rewrite.
Before the next rewrite
Start with a four-scene example. No account required.