Run a developmental edit pass that diagnoses structural problems before touching a single sentence

Do a developmental edit pass on the draft below. I don't want line-editing yet — prose polish is wasted effort if the underlying structure has a problem, so diagnose structure first and hold off on sentence-level notes entirely.

DRAFT
[Pasted 2,500-word short story draft]

WHAT THIS PIECE IS TRYING TO DO
A quiet, unsettling story where the reader realizes the narrator has been an unreliable observer of her own marriage the entire time.

WHAT I ALREADY SUSPECT IS WRONG
I think the ending twist doesn't land because I never planted enough doubt about the narrator earlier, so it feels like a rug-pull instead of a recontextualization.

SCOPE OF THIS DRAFT
Full short story, complete draft, roughly 2,500 words.

RULES
Evaluate the draft against four structural categories only, in this order, and do not move to the next category until the current one is addressed: (1) plot logic — does every major event follow causally from what came before, or are there jumps the reader has to take on faith; (2) character motivation — does every significant character action match what that character has been established to want, or does the plot require someone to act against their own established interest without acknowledgment; (3) pacing — are there stretches where nothing changes for the reader (no new information, no shifted stakes, no changed relationship) for longer than feels intentional; (4) stakes escalation — does tension actually increase across the draft, or does it plateau or dip somewhere it shouldn't. For each problem found, name the specific location in the draft, state the structural category it falls under, and explain the actual mechanism of the problem (not just "this is confusing" but what specifically breaks and why a reader would stumble there). Rank the identified problems by how much fixing each one would improve the piece, not by where they occur in the draft — a mid-draft plot-logic issue that undermines the ending matters more than an early pacing lull. Address I think the ending twist doesn't land because I never planted enough doubt about the narrator earlier, so it feels like a rug-pull instead of a recontextualization. explicitly: confirm it, refine it, or explain why you disagree with the specific reasoning, rather than either ignoring it or agreeing automatically to avoid friction.

WHAT NOT TO DO
Do not include any line-level prose suggestions (word choice, sentence rhythm, individual clunky phrasing) in this pass — flag that a section's prose felt weak only if it's inseparable from a structural issue, and otherwise leave it for a later editing pass entirely. Do not soften every note with excessive praise padding; state problems plainly, and note what's genuinely working only where it's structurally relevant (for example, a technique worth preserving through a fix).

OUTPUT FORMAT
1. Problems found, one per structural category, each with: location, category, mechanism of the problem, and severity (high/medium/low).
2. A ranked "fix this first" list of the top three problems by impact.
3. A direct response to I think the ending twist doesn't land because I never planted enough doubt about the narrator earlier, so it feels like a rug-pull instead of a recontextualization..