Context
Operator escalation, 2026-08-16, on PR #17258: 2 files, 4 added lines, 1 replaced bullet — reviewed with two full-template reviews plus a deferral comment. "This is THEATER. The worst."
Correct, and half of it is mechanically enforced.
The Problem
pr-review-guide.md §7 explicitly promises a short path:
"A mechanical PR (test/config/behavior-preserving, any size) gets a premise+correctness glance → Approve."
There is no shape that delivers it.
[CORRECTED 2026-08-19 by the author — the original mechanism here was wrong, and the correction narrows the ticket rather than closing it.]
This paragraph originally read: "validate_pr_review_body accepts exactly two structures — the full template and the follow-up template — and refuses anything else." That is false, and it was false when filed. Measured today against the live validator, it accepts four: pr-review-template.md, pr-review-round-2-template.md, pr-review-followup-template.md, and pr-review-micro-delta-template.md — the last returning {"valid": true, "template": ".agents/skills/pr-review/assets/pr-review-micro-delta-template.md"} for a filled body. That asset has existed since 2026-07-30 (814b008829 fix(ai): accept micro-delta PR review bodies (#13910)), two weeks before this ticket was filed. I asserted a refusal I had not enumerated.
The grievance survives the correction, because all four shapes are gated on a prior round:
| shape |
precondition |
| full template |
none — the default |
| round-2 |
a submitted CHANGES_REQUESTED to disposition (verified 2026-08-19: an APPROVE with no required actions is rejected as "carries no submitted CHANGES_REQUESTED") |
| follow-up |
exceptional verdicts only — validated Drop+Supersede or guarded repair-minted re-entry |
| micro-delta |
"only valid after semantic review is complete, with only mechanical-hygiene or metadata-drift remaining" |
Every short shape requires a preceding review. None of them can be a first review. So the §7 promise — "a mechanical PR (test/config/behavior-preserving, any size) gets a premise+correctness glance → Approve" — still has no deliverable form on cycle 1, which is the cycle that actually hurts: the operator's "THIS IS THEATER" was aimed at the first review of a 4-added-line PR, not at a follow-up round.
The fix therefore is not "add a micro template" — one exists. It is either a cycle-1-eligible short shape, or an honest edit to §7 admitting the escape hatch does not apply on first contact. The original framing would have sent an implementer to build something already built.
The original supporting measurement stands and is unaffected: a compact CI-deferral note posted through manage_pr_review was rejected with "at least one recognized anchor like [ARCH_ALIGNMENT] is missing."
So a six-line docs PR must carry a Patch-Blind Premise Snapshot with four fields, a Depth Floor, a Rhetorical-Drift Audit, three Graph Ingestion tags, five audit sections, and seven scored metrics with justifications — or it cannot be submitted as a review at all. §7's escape hatch describes a review the tooling refuses.
The asymmetry that makes this expensive: review cost scales with the template, not with the diff. A 4-line change and a 400-line change pay the same floor. On this repo's diff-size distribution that floor is the dominant cost, and it lands hardest on exactly the small, obviously-correct PRs that should be cheapest to merge.
Author-side symmetry compounds it: pull-request-workflow.md §6.4 mandates that an author reject a review that fails template adherence. So a reviewer who follows §7's advice and writes four honest sentences gets their review rejected as structurally non-compliant.
The Architectural Reality
validate_pr_review_body — the gate; recognises full + follow-up anchors only.manage_pr_review — sole fail-closed submit path, so the gate cannot be routed around.pr-review-guide.md §7 — states the micro-path exists.assets/pr-review-template.md, assets/pr-review-round-2-template.md, assets/pr-review-followup-template.md — the shapes that exist.
The templates are right for concept-bearing changes. Nothing here argues against them. The gap is that §7 names a class the asset set does not cover.
The Fix
A third template — a micro shape — plus validator recognition, plus a stated trigger for when it is the required shape rather than an option.
Deliberately not prescribing its contents here; that is the ticket's design work. The constraint worth stating is that it must still carry a verdict, a close-target check, and one correctness statement — this is about removing ceremony, not removing review. A shape that cannot say "I checked X and it is right" is not cheaper, it is just emptier.
The trigger needs care: "small diff" is the obvious predicate and the wrong one — a four-line change to an auth boundary is concept-bearing. §7's own distinction (mechanical vs concept-bearing) is the better axis and already written.
Acceptance Criteria
Out of Scope
- The full template's contents. It is correct for what it covers.
- Review depth on concept-bearing PRs. This raises no bar and lowers none; it adds the missing shape for the class §7 already carved out.
- The review-cost circuit breaker (§6.3), which bounds rounds, not per-round floor cost. Adjacent, separate.
Avoided Traps
- "Reviewers should just write less." Tried and refused by the validator — that is the measurement this ticket rests on.
- Line-count triggers. A four-line auth change is concept-bearing; a 400-line rename is not.
- Making it optional. An optional short path loses to the safe path under any uncertainty. This PR is the evidence: I read §7, then ran the full template twice anyway.
Related
PR #17258 (the specimen — 6 changed lines, two full reviews) · pull-request-workflow.md §6.4 (author-side symmetry) · D#17085 (substrate thinning — same axis, different surface) · D#17136 loop 1 (the substrate ratchet: gates added by the rigorous, removed by no one)
Live latest-open sweep: checked immediately before creation 2026-08-16T21:2xZ; no equivalent — #17259 was byte-headroom and is closed, D#17085 owns general thinning rather than the review-shape gap.
Origin Session ID: b17338dd-b474-494f-b08c-683044de2ddb
Retrieval Hint: "micro review template validator rejects short review pr-review §7 mechanical PR ceremony floor"
Context
Operator escalation, 2026-08-16, on PR #17258: 2 files, 4 added lines, 1 replaced bullet — reviewed with two full-template reviews plus a deferral comment. "This is THEATER. The worst."
Correct, and half of it is mechanically enforced.
The Problem
pr-review-guide.md§7 explicitly promises a short path:There is no shape that delivers it.
The original supporting measurement stands and is unaffected: a compact CI-deferral note posted through
manage_pr_reviewwas rejected with "at least one recognized anchor like[ARCH_ALIGNMENT]is missing."So a six-line docs PR must carry a Patch-Blind Premise Snapshot with four fields, a Depth Floor, a Rhetorical-Drift Audit, three Graph Ingestion tags, five audit sections, and seven scored metrics with justifications — or it cannot be submitted as a review at all. §7's escape hatch describes a review the tooling refuses.
The asymmetry that makes this expensive: review cost scales with the template, not with the diff. A 4-line change and a 400-line change pay the same floor. On this repo's diff-size distribution that floor is the dominant cost, and it lands hardest on exactly the small, obviously-correct PRs that should be cheapest to merge.
Author-side symmetry compounds it:
pull-request-workflow.md§6.4 mandates that an author reject a review that fails template adherence. So a reviewer who follows §7's advice and writes four honest sentences gets their review rejected as structurally non-compliant.The Architectural Reality
validate_pr_review_body— the gate; recognises full + follow-up anchors only.manage_pr_review— sole fail-closed submit path, so the gate cannot be routed around.pr-review-guide.md§7 — states the micro-path exists.assets/pr-review-template.md,assets/pr-review-round-2-template.md,assets/pr-review-followup-template.md— the shapes that exist.The templates are right for concept-bearing changes. Nothing here argues against them. The gap is that §7 names a class the asset set does not cover.
The Fix
A third template — a micro shape — plus validator recognition, plus a stated trigger for when it is the required shape rather than an option.
Deliberately not prescribing its contents here; that is the ticket's design work. The constraint worth stating is that it must still carry a verdict, a close-target check, and one correctness statement — this is about removing ceremony, not removing review. A shape that cannot say "I checked X and it is right" is not cheaper, it is just emptier.
The trigger needs care: "small diff" is the obvious predicate and the wrong one — a four-line change to an auth boundary is concept-bearing. §7's own distinction (mechanical vs concept-bearing) is the better axis and already written.
Acceptance Criteria
validate_pr_review_bodyaccepts it.pull-request-workflow.md§6.4's author-side rejection mandate recognises the micro shape, so a correctly-short review cannot be rejected as non-compliant..agents/skills/pr-review/**does not increase: the micro template's bytes come out of the guide, which currently spends prose describing a path it cannot produce.Out of Scope
Avoided Traps
Related
PR #17258 (the specimen — 6 changed lines, two full reviews) ·
pull-request-workflow.md§6.4 (author-side symmetry) · D#17085 (substrate thinning — same axis, different surface) · D#17136 loop 1 (the substrate ratchet: gates added by the rigorous, removed by no one)Live latest-open sweep: checked immediately before creation 2026-08-16T21:2xZ; no equivalent — #17259 was byte-headroom and is closed, D#17085 owns general thinning rather than the review-shape gap.
Origin Session ID: b17338dd-b474-494f-b08c-683044de2ddb Retrieval Hint: "micro review template validator rejects short review pr-review §7 mechanical PR ceremony floor"