Studio → a project → Clues.
The Prop Bible already keeps the locket looking like the same locket. This tracks whether it matters.
Those are different failures. A clue can be rendered perfectly in every one of its eleven shots and still be a dead plot thread — and no amount of looking at pixels will tell you.
### What gets caught
- Orphaned setup — planted in scene 3 and never seen again. Every frame of it is correct. The audience still noticed.
- Unplanted payoff — the scar identifies the twin in scene 41, but was never established. It reads as a cheat rather than a reveal.
- A gap — planted and paid off, but missing from the scenes in between where the story needs it present.
- On the plan but not on the screen — marked as set up in scene 3, but it does not appear in any shot there.
### Flagging one
Give it a name, the scenes where it is planted and paid off, and optionally the scenes where it must stay visible. Weight it: colour, supporting, or load-bearing. A load-bearing clue that does not pay off is a red audit flag.
Link it to a Bible entry where you can. A linked prop is checked against the actual shot links, which is far more reliable than searching prose. For a spoken clue, make the name the exact phrase — that is what is searched for.
### Suggestions
The page suggests candidates from what is already in the project. A prop appearing in exactly one scene is the classic orphaned setup: you put it there for a reason and the reason never arrived.
### On scene numbers
Clues reference scene ORDER rather than scene id, because scene rows are re-created every time the script is re-parsed and ids would dangle silently. Inserting a scene can shift a clue by one — the audit reports scene numbers that no longer exist rather than ignoring them.