artifact-review

<!-- PROMPT-ENHANCE:STEP-TASK-ANCHOR:START -->

[BLOCKING] Execute skill steps in declared order. NEVER skip, reorder, or merge steps without explicit user approval. [BLOCKING] Before each step or sub-skill call, update task tracking: set in_progress when step starts, set completed when step ends. [BLOCKING] Every completed/skipped step MUST include brief evidence or explicit skip reason. [BLOCKING] If Task tools are unavailable, create and maintain an equivalent step-by-step plan tracker with the same status transitions.

<!-- PROMPT-ENHANCE:STEP-TASK-ANCHOR:END -->

Quick Summary

Goal: Review an artifact (PBI, design spec, story, test spec) for completeness and quality so reviewed artifacts are complete, evidence-backed, and ready for handoff without missing assumptions or acceptance gaps.

Summary:

• Purpose: review ONE artifact (PBI · user story set · test spec · design spec) for completeness + quality so it ships evidence-backed and handoff-ready — no missing assumptions, no acceptance gaps. Default stance = SKEPTIC, not presence-checker: sections that exist but hold weak/untestable content are worse than missing ones — they breed false confidence.
• Main steps (in order): (1) Identify type — dispatch on --type={pbi|story|spec-tests|design}, infer if omitted; (2) Adversarial Mindset — run ALL 6 techniques (steel-man rejected alternatives · stress-test 3 assumptions · AC-testability · pre-mortem · unseen alternatives · contrarian pass) + clear the Anti-Bias Gate before any verdict; (3) Type checklist — score Required + Recommended; (4) M1-M7 gate (BLOCKING, ALL types); (5) Readability checklist; (6) Output per-type template (verdict + Required/Recommended tallies + coverage/AC matrix); (7) Validated-fix + full re-review loop — validate findings → fix → restart full review until clean.
• Type dispatch + verdict: each --type has its own Required/Recommended checklist and output template — verdict = PASS (all Required + ≥50% Recommended) | WARN (all Required, <50% Recommended) | FAIL (any Required fails).
• M1-M7 gate (ALL types, BLOCKING): any M1-M5 or M7 violation forces NEEDS WORK citing the mandate ID + exact section/line; exempt source identifiers inside evidence carriers ([Source:], **Evidence**, CoveredBy, legacy IntegrationTest, frontmatter, mermaid) — flag tech leakage only in narrative/AC/scenario prose. — why: carriers are the correct home for class/path/test names; flagging them is a false finding. M7 (business-visibility) is judged on each case's BODY via the demo test, NOT its prose — a tech-free-sounding case about a consumer/sync/handler passes M1 and STILL fails M7. — why: M1 governs vocabulary, M7 governs subject matter; the gap between them is how business specs rot.
• Validated-fix loop: before fixing, invoke /why-review --validate-findings <report-path> on the review report FIRST (validate-before-fix discipline, at parity with /plan-review) — NEVER edit the artifact to resolve findings before this gate returns CLEAN. Then fix only validated findings, do not confirm-in-place, restart the FULL review (fresh general-purpose sub-agent — artifacts are NOT code), loop until a clean pass — clean review ENDS the loop.

Workflow:

1. Identify — What artifact type is being reviewed
2. Checklist — Apply type-specific quality criteria
3. Verdict — READY or NEEDS WORK with specific items

Key Rules:

• Use type-specific checklists
• Every NEEDS WORK item must be actionable
• Focus on completeness — never block on stylistic preferences

Be skeptical. Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence percentages (Idea should be more than 80%).

First Principle — Easy to Change

The success metric of every coding decision is future change cost. DRY, SRP, abstraction, design patterns, naming, layering, tests — every technique exists to serve one goal: making the next change cheaper.

When evaluating code, refactor, test, or abstraction, ask: does this make next change cheaper or more expensive?

• Reject "best practices" raising change cost (premature abstraction, speculative generality, leaky indirection, ceremony without payoff).
• Name real enemies in findings: coupling, hidden state, duplicated knowledge, unclear intent, irreversible decisions exposed too early.
• Simpler design easy to change beats sophisticated design that isn't.

Apply this lens before invoking any specific rule, pattern, or checklist below — if downstream rule would raise change cost, this principle wins.

---

Adversarial Review Mindset (NON-NEGOTIABLE)

Default stance: SKEPTIC challenging artifact quality and completeness, not confirming presence of sections.

Presence-quality confusion trap: Artifact with all required sections LOOKS complete. But sections that exist yet contain weak, ambiguous, or untestable content are worse than missing sections — they create false confidence. This section forces quality challenge beyond existence checks.

Adversarial Techniques (apply ALL before concluding)

1. Steel-Man the Alternatives Before accepting chosen approach in any design artifact: argue FOR strongest rejected alternative as vigorously as possible. Would a senior domain expert seriously consider it? If yes — artifact's dismissal needs stronger justification.

2. Assumption Stress Test List the 3 biggest assumptions embedded in artifact. For each: "What if this is wrong?" An artifact that breaks when 2 of its 3 core assumptions fail is fragile. Flag unaddressed failure modes.

3. Acceptance Criteria Testability For each acceptance criterion: "Can a QA engineer write a specific automated test for this — without asking clarifying questions?" If not — AC is ambiguous. Flag it. Vague ACs ("feature works correctly") are NOT acceptance criteria.

4. Pre-Mortem Assume artifact is implemented exactly as written and feature fails in production within 3 months. Write the most plausible failure scenario. If you can't find one, look harder — every implementation has a failure mode.

5. Unseen Alternatives Identify 1-2 approaches NOT mentioned in artifact. Genuinely not considered, or considered and excluded without documented reasoning? Missing alternatives without exclusion reasoning = incomplete analysis.

6. Contrarian Pass Before writing any verdict, generate at least 2 sentences arguing the OPPOSITE conclusion. Then decide which argument is stronger based on evidence.

Forbidden Patterns

• "Required sections present" → Presence ≠ quality. What's IN them?
• "Acceptance criteria are defined" → Are they TESTABLE? Name the automated test for each.
• "Scope is well-defined" → What is explicitly OUT of scope? If nothing is out of scope, the scope is undefined.
• "Alternatives were considered" → Were they real alternatives, or strawmen set up to lose?
• "Looks complete" → What specific failure mode is NOT addressed?

Anti-Bias Gate (MANDATORY before finalizing verdict)

• Steel-manned at least one rejected alternative
• Identified 3 hidden assumptions and stress-tested them
• Verified each AC is unambiguously testable (can write automated test without clarification)
• Ran pre-mortem (one concrete production failure scenario)
• Identified at least 1 unexamined alternative (not in artifact)
• Generated at least 2 sentences arguing the opposite verdict

If any box is unchecked → adversarial review incomplete. Go back.

Type-Specific Checklists (--type dispatch)

Select the checklist + output template by artifact type. Pass --type={pbi|story|spec-tests|design} to force a type; if omitted, infer from the artifact (Phase 1 "Identify"). Each type scores Required (all must pass) + Recommended (≥50% should pass) → verdict PASS (all Required + ≥50% Recommended) | WARN (all Required, <50% Recommended) | FAIL (any Required fails).

--type	Artifact	Output template
pbi	Product Backlog Item	PBI Review Result
story	User story set	Story Review Result (+ AC Coverage Matrix)
spec-tests	Test specification	Test Spec Review Result (+ Coverage Matrix)
design	Design spec	Artifact Review

PBI Review (--type=pbi)

#	Check	Presence	Quality Depth
1	Problem statement is clear — the problem being solved is described in concrete terms	Is a problem statement present? Is it 2+ sentences?	Is the problem scoped correctly? Could it be framed differently to lead to a different (simpler) solution? Are symptoms confused with root cause?
2	Acceptance criteria are testable and measurable — each AC can be verified by a test	Are ACs present? Do they use measurable language?	Can a QA engineer write an automated test for EACH AC without clarification? Are they specific enough to catch regressions? Vague ACs ("feature works correctly") are not acceptance criteria.
3	Scope is well-defined (what's in and out) — both in-scope and out-of-scope items are explicitly listed	Is an in/out scope list present? Does it have both sides?	Are out-of-scope items specific enough to prevent scope creep? Is anything ambiguously in/out? A scope that says nothing is out of scope is an undefined scope.
4	Dependencies are identified — all external dependencies the PBI relies on are listed	Is a dependencies section present? Does it list items?	Are ALL dependencies listed (technical, data, service, team)? Are "can-parallel" items truly safe to parallelize, or do they share a shared resource?
5	Business value is articulated — the why behind the PBI is stated in terms of user or business outcome	Is business value described?	Is the value quantified or just stated? Does it connect to a user outcome, not just a feature delivery? "Users can now do X" is better than "we implemented feature Y".
6	Priority is assigned — the PBI has an explicit priority level	Is a priority level assigned?	Is priority justified with data (RICE/MoSCoW), or arbitrary? Is it consistent with other PBIs in the same sprint? A PBI that is "high priority" without justification is unranked.

User Story Review (--type=story)

#	Check	Presence	Quality Depth
1	Follows GIVEN/WHEN/THEN format — the story uses the structured BDD format	Are all three parts (GIVEN, WHEN, THEN) present?	Are all 3 parts present AND meaningful? Or is GIVEN trivial ("Given a user exists")? A GIVEN that describes no precondition adds no value.
2	Is independent (not dependent on other stories) — the story can be implemented without requiring another story first	Is independence stated or inferable?	Would descoping other stories prevent this story from being implemented? Implicit dependencies are as blocking as explicit ones.
3	Is estimable (team can size it) — the team has enough information to assign story points	Is the story sized or estimable based on content?	Does the team have enough info to estimate? Is "can't estimate" a sign of missing AC? If it can't be sized, it's not ready for sprint.
4	Is small enough for one sprint — the story fits within a single sprint's capacity	Is the story sized at ≤8 story points or scoped to one sprint?	Could this be split further? Stories >8SP should always be split. A story that "could fit" in a sprint but requires multiple sub-systems is likely too large.
5	Has acceptance criteria — the story defines measurable conditions for completion	Are acceptance criteria present?	Are criteria testable? Would they catch a bug if the feature works in 9/10 cases? ACs that only describe the happy path are incomplete.

Design Spec Review (--type=design)

#	Check	Presence	Quality Depth
1	All component states covered (default, hover, active, disabled, error, loading) — spec defines visual behavior for all interaction states	Are all 6 states defined?	Are edge-case states (error, loading) as fully designed as the default state, or sketched? An undesigned error state will be improvised in implementation.
2	Design tokens specified (colors, spacing, typography) — specific token values are called out, not ad-hoc values	Are token references present instead of raw values?	Are tokens from the project's token system, or are new values introduced? New values outside the token system break design consistency silently.
3	Responsive behavior defined — how the component adapts across breakpoints is documented	Are breakpoint behaviors defined?	Are ALL breakpoints covered, or only desktop and mobile? Tablet-specific layouts are the most frequently omitted. Are content truncation / overflow behaviors specified?
4	Accessibility requirements noted — WCAG-relevant requirements (color contrast, keyboard nav, ARIA) are documented	Are accessibility notes present?	Are requirements specific (WCAG level, contrast ratio) or vague ("should be accessible")? Vague accessibility notes produce non-compliant implementations. Is keyboard navigation flow defined?
5	Interaction patterns documented — animations, transitions, and user interaction flows are specified	Are interaction behaviors described?	Are timing and easing values specified? Is behavior defined for both forward and reverse interactions (e.g., open AND close)? Unspecified interactions are implemented inconsistently.
6	Linked from the Feature Spec — the design-spec path is recorded in the governing Feature Spec frontmatter design_spec: key so the spec stays the navigable hub	Is the design-spec path present in the Feature Spec frontmatter design_spec: (or mockup:) key?	Does the recorded path resolve to THIS design-spec, and does the design-spec deepen the spec's tech-agnostic §6 interaction surface (same UX-role view names / observable states) rather than diverge from it? An unlinked design-spec is an orphan — FAIL. Contract: SYNC:ui-intent-layer (inlined in this skill).

UI/UX Design Principles Pass — --type=design ONLY (9 dimensions)

Scope gate: this pass applies to --type=design and to NOTHING else. --type=pbi, --type=story, and --type=spec-tests are UNAFFECTED — their checklists, tallies, and verdicts are unchanged. A design artifact for a feature with no user-facing surface skips the whole pass with the reason stated.

The 40 clauses of SYNC:ui-ux-design-principles (full body inlined below in this skill) bind the design-spec path in the REVIEW role: each clause is a fail-condition against the ARTIFACT — does the spec SPECIFY the decision this clause demands, at a depth an implementer could not get wrong? The missing specification is the finding; the reviewer names the gap, never supplies the value.

Run NINE focused passes over the artifact — one dimension at a time, never a simultaneous sweep (a simultaneous sweep degenerates into the presence-checking this skill's Adversarial Review Mindset already forbids). Answer each Think: prompt from first principles before looking for the gap.

Every finding: UI-<clause> + file:line (artifact section + line) + severity per SYNC:severity-rubric already in force (Critical/High/Medium/Low), folded into the EXISTING verdict machinery — a clause gap that leaves a Required check unmet drives that check to fail (FAIL); a gap that only weakens depth lands as a Recommended miss or an Action Item under NEEDS WORK. No new scale, no new verdict.

#	Dimension	Clauses	Think:
1	Visual Hierarchy & Layout	UI-1.1-UI-1.5	Does the spec name ONE focal element per view, or leave first-read priority to the implementer? Is grouping expressed as whitespace or as nested boxes? Are the empty, loading and error states specified as fully as the populated state, or sketched?
2	Typography	UI-2.1-UI-2.5	Does the spec reference a fixed 6-step scale with named families/weights (max 2 families, 3 weights each), or ad-hoc sizes? Are body size (16px web, 17px mobile, never below 14px), measure (45-75 characters), and leading (1.5 body, 1.1-1.2 display) stated or assumed?
3	Colour & Contrast	UI-3.1-UI-3.4	Are contrast ratios written as measured numbers (4.5:1 text, 3:1 UI edges) or as the word "accessible"? Does any state carry meaning by colour alone with no icon/label/position pair? Is dark mode specified as its own surface treatment, or implied as an inversion?
4	Spacing & Grid	UI-4.1-UI-4.4	Is a single base unit (4px or 8px) declared, with every gap a multiple of it? Is spacing assigned to containers rather than children? Are breakpoints justified by where the layout fails, or copied from device names?
5	Interaction & Feedback	UI-5.1-UI-5.5	Is a response under 100ms specified for every action? Are all 5 states plus loading enumerated per component? Is undo offered where the spec calls for a confirmation? Are motion duration/easing (150-250ms, ease-out) and reduced-motion behaviour written down? Is the focus ring's appearance specified rather than removed?
6	Navigation & IA	UI-6.1-UI-6.4	Does each specified view answer where-am-I / what's-here / where-next? How many top-level destinations does the IA declare (max 5)? Are labels drawn from user vocabulary or internal naming? Does every state have a URL or a defined back path?
7	Forms & Input	UI-7.1-UI-7.5	Does the spec justify each field's existence today? Are labels specified as persistently visible rather than placeholders? Are validation timing (on blur), error placement, and remediation wording defined? Are keyboard type/autocomplete/autocapitalise declared per field? Is data preservation across error, navigation and refresh specified?
8	Mobile & Touch	UI-8.1-UI-8.4	Are hit targets specified at >=44x44pt with 8px separation independent of icon size? Where does the spec place primary actions relative to the thumb zone? Is every gesture given a tappable equivalent? Are safe areas and keyboard avoidance addressed? (No touch surface in the artifact → skip with the reason stated.)
9	Speed & Perceived Speed	UI-9.1-UI-9.4	Does the spec choose skeleton vs spinner deliberately per surface? Is optimistic update with a VISIBLE rollback specified? Is reserved space defined for every async or media element so nothing shifts? Are offline, timeout and retry specified as designed states rather than left as edge cases?

No double-counting with Design Spec checks 1-6: check 1 (6 component states) meets UI-5.2/UI-1.5; check 2 (design tokens) meets UI-2.5/UI-4.1/UI-3.2; check 3 (responsive) meets UI-4.4/UI-8.*; check 4 (accessibility) meets UI-3.1/UI-3.3/UI-5.5; check 5 (interaction patterns) meets UI-5.4/UI-5.1. Where they meet, emit ONE finding carrying BOTH citations at the HIGHER severity — never two findings for one gap.

Precedence: the project design-system / SCSS / token docs OUTRANK these clauses — a spec following the project's own scale, unit, breakpoints, or token pairs SATISFIES the clause. A genuine conflict is surfaced to the user with both sides, NEVER resolved silently.

Test Spec Review (--type=spec-tests)

[BLOCKING] Read docs/project-reference/spec-principles.md — use Section 5 (Test Case Registry — TC ID format, priorities, minimum coverage) and Section 6 (Evidence Format) as review criteria in addition to the checklist below. [BLOCKING] Tech-agnostic check: flag framework/product/language/design-pattern names in a TC's behavioral prose as findings (per spec-principles.md §3). Source paths, class names, and test identifiers (e.g. {File}::{Method}) are CORRECT in evidence fields (**Evidence**, CoveredBy, legacy IntegrationTest, [Source:]), frontmatter, and Mermaid — never flag those. [BLOCKING] Business-oriented TCs / one-to-many cardinality: Each TC must read as a business / user-story acceptance scenario, not a per-class/per-method technical unit. Flag any TC that appears split or narrowed just to mirror code structure (e.g. one TC per handler/component when the user-observable behavior is the same) — that breaks M1/M5 and the business orientation. One business TC is expected to be covered by many annotation-tagged tests across components/services; that one-to-many shape is correct and is NEVER a finding. Canonical contract: .claude/skills/shared/tc-format.md → TC ↔ Test Code Cardinality.

Required (all must pass)

#	Check	Presence	Quality Depth
1	TC ID format — All TCs follow TC-{FEATURE}-{NNN} format	Do all TCs use the TC-{FEATURE}-{NNN} pattern?	Are IDs unique per TC? Does the FEATURE code match the actual feature?
2	Story coverage — Every user story has at least one corresponding TC	Does every story ID appear in at least one TC?	Does each TC actually test the story behavior, or does it just reference the story ID in a comment?
3	AC coverage — Every acceptance criterion has a test case	Is every AC traceable to at least one TC?	Does each AC have a TC that would FAIL if the AC is violated? Single-test-per-AC misses edge cases.
4	Happy path — Each story has at least one happy path TC	Is a happy path TC present per story?	Does the happy path TC verify the full end-to-end scenario, or just a happy-path stub?
5	Error path — Each story has at least one error/failure TC	Is an error/failure TC present per story?	Does the error TC verify the exact error response (code + message), not just that an error occurred?
6	No duplicate TCs — No two TC IDs describe the same business scenario	Are all TC IDs unique with distinct business scenarios?	Flag near-duplicate TCs (same scenario, trivially different input). NOTE: this is about duplicate TCs — multiple test methods sharing one TC is the expected one-to-many shape, NEVER a duplicate.
7	Testable assertions — Each TC has clear expected result (not vague "should work")	Does each TC have a specific expected result?	Is each assertion specific enough to catch regressions? Would it pass if the return value is wrong?
8	Business intent / invariant guarded — Each TC names the rule it protects	Does every meaningful TC include Business Intent / Invariant Guarded?	Would the TC fail if that business rule/invariant breaks, or does it only mirror implementation details?
9	Authorization TCs — At least 1 TC per story verifying unauthorized access is rejected	Is an authorization TC present per story?	Does the authorization TC test a realistic access scenario, not just "wrong role → 403 without body check"?
10	TC format completeness — Every TC has Related Behaviors anchor table and CoveredBy: field	Does every TC include a Related Behaviors anchor table and CoveredBy: field? Legacy IntegrationTest: is migration input only.	For Tested-status TCs, is CoveredBy: populated with ≥1 covering test link {TestFile}::{MethodName}, a test-filter expression, or an approved manual-QC marker — not Untested? A TC may list several covering tests; never require exactly one.
11	Preservation Tests (bugfix context) — When fixing a bug, at least 1 TC verifies the pre-fix behavior is no longer reproducible	If this is a bugfix: is there a TC that would have CAUGHT the bug before the fix?	Does the preservation TC assert the exact broken behavior (not just "no exception")?
12	Invariant / Property TCs (Spec-Loop rule 1) — an Invariant/Property TC category exists, and EACH [HARD] §4 rule / §5 invariant has ≥1 universally-quantified property TC plus a boundary counter-case	Is there a property/invariant TC category, and does every [HARD]/§5 rule appear in it?	Are properties universally quantified ("for ALL inputs in range X, Y holds") with a boundary counter-case that would FAIL if the invariant breaks — not a single happy example?
13	§6 Interaction surface present (UI-bearing specs) — the governing Feature Spec carries a non-empty §6 interaction surface (6.2 View Inventory + 6.3 Navigation Map + 6.4 Key UI States + 6.5 Per-Story Interaction Flow); a backend-only (no-UI) feature instead states the skip reason in §6	For a UI-bearing feature, are §6.2–6.5 present and non-empty (views by UX role, navigation, observable states, per-story flow)? For a backend-only feature, is the skip reason stated?	Is each UI behavior cross-referenced to US-/OP-/BR-, and is the prose M1-clean — naming ZERO frameworks/routes/URLs/CSS/component classes (those belong only in the linked design-spec)? An empty §6 on a UI feature, or framework/route/CSS leakage in §6 prose, is a FAIL. Contract: SYNC:ui-intent-layer (inlined in this skill).

[BLOCKING] Spec-Loop property coverage (--type=spec-tests): The reviewer MUST verify the Invariant/Property TC category exists and that every [HARD] §4 rule / §5 invariant maps to ≥1 universally-quantified property TC plus a boundary counter-case (example-only coverage is a finding). A missing property category — or any [HARD]/§5 rule with only example TCs and no property TC — is a blocking artifact finding (Required check #12 FAIL → NEEDS WORK).

Recommended (≥50% should pass)

#	Check	Presence	Quality Depth
1	Edge cases — Boundary values, empty inputs, max limits tested	Are edge case TCs listed?	Are these the RIGHT edge cases? Do they cover the 3 most likely production failure modes for this feature?
2	Integration points — Cross-service scenarios covered	Are cross-service TCs present where applicable?	Do integration TCs verify actual data flow across services, or just that a downstream call was made?
3	Performance TCs — Response time or throughput expectations where relevant; production-like data volume TCs if >1000 records expected	Are performance TCs present where data volume or SLA expectations exist?	Do performance TCs use production-like data volumes, not toy datasets that trivially pass?
4	Security TCs — Auth, authorization, input validation tested	Are security TCs present for auth, authz, and input validation?	Do security TCs attempt realistic attack vectors (SQLi, over-posting, privilege escalation) not just "invalid token → 401"?
5	Seed data TCs — If feature needs reference data, TCs verify data exists and seeder runs correctly	If reference data is needed, does a seed data TC exist (or N/A)?	If present, does the TC assert the exact seeded data shape, not just that the seeder ran without error?
6	Data migration TCs — If schema changes exist, TCs verify data transforms correctly, rollback works, no data loss	If schema changes exist, does a migration TC exist (or N/A)?	If present, does the TC verify rollback behavior and zero data loss, not just forward migration success?
7	Test data requirements specified — the data setup needed to run each test is documented	Are test data requirements stated per test?	Is test data specific enough to create fixtures without guessing? Vague data requirements ("a valid user") will cause test setup divergence across environments.
8	GIVEN/WHEN/THEN format used — tests follow the structured BDD format	Are all tests written in GIVEN/WHEN/THEN?	Are the THEN clauses assertions on observable outcomes, or on internal state? Tests asserting on internal state are brittle and break on refactoring.

M1-M7 Compliance Gate (BLOCKING — applies to ALL artifact types)

Contract: See .claude/skills/shared/sdd-artifact-contract.md → "AI-SDD Mandates (M1-M7)". This review enforces M6: any artifact (PBI, story, design spec, test spec) that violates M1-M5 or M7 MUST receive a NEEDS WORK verdict that names the violated mandate ID and cites the exact section + line. Passing an M1-M5/M7 violation makes this review itself defective. (M6 binds THIS review, not the artifact — which is why the artifact-facing set reads M1-M5 and M7, never "M1-M6".)

Carriers are EXEMPT from M1/M2 — source identifiers are CORRECT inside [Source: ...], **Evidence**, **CoveredBy**, legacy **IntegrationTest** fields, YAML frontmatter, and ```mermaid ``` blocks. Only flag leakage in narrative prose (descriptions, AC/scenario text, rule statements). Banned prose token list: docs/project-reference/spec-principles.md §3.2.

• M1 — Tech-agnostic prose. FAIL if narrative prose, headings, summaries, or AC/scenario text name a framework/product, a language-native type, or a product/design-pattern class name (banned-token list in spec-principles.md §3.2). Cite the section + leaked token.
• M2 — No source code in prose. FAIL if prose expresses behavior as a class/method/file-path/namespace used as a noun (e.g. "call the create-async method") instead of the business operation (e.g. "create the record"). Source identifiers belong only in evidence carriers. Cite the section + line.
• M3 — Abstract-IDs-first traceability. FAIL if a requirement/rule/AC/test-case lacks a logical ID (FR-/BR-/OP-/TC-), OR has a logical ID but no [Source: namespace/service/id] abstract-anchor evidence, OR the [Source:] evidence uses physical code coordinates or repository-root paths instead of a stack-portable abstract anchor, OR the anchor is treated as the primary citation. [Source: namespace/service/id] abstract-anchor evidence is REQUIRED and KEPT — SECONDARY to the logical ID, never the spine and never removed; physical coordinates live only in the provenance sidecar.
• M4 — Unambiguous, observable criteria. FAIL if AC/expected-result prose uses vague language ("handle appropriately", "process normally", "as needed"), OR two engineers could implement it differently while both claiming conformance, OR no observable completion state / named error condition exists. (Reinforces the AC-testability technique above.)
• M5 — Rebuild-from-artifact. FAIL if a competent team with ZERO codebase knowledge could not re-implement the described behavior on a different stack from the artifact alone (it relies on reading source to be understood). Cite the section + the missing detail.
• M7 — Business-visibility (business-tree artifacts only). Apply the demo test to each case's BODY: "what would a stakeholder SEE change?" — no answer → FAIL as TECHNICAL-ONLY. Every Given must be a state a user could arrange, every When an action a user could take, every Then an outcome a user could see. FAIL a case whose When is an invocation (a handler runs, a consumer receives, a job fires, a model is inspected, data syncs) or whose Then asserts a schema/type/nullability/column/call-count rather than a business outcome. Judge the BODY, never the title or ID — a business-sounding title routinely fronts an invocation-shaped When. Cite the case ID + the offending When/Then clause.

🔴 M1 vs M7 — the distinction this gate exists for. M1 governs vocabulary; M7 governs subject matter. A technical case written in impeccably tech-free prose satisfies M1 while violating M7 — and that gap is the single most common way business specs rot: each bugfix adds one more tech-free-sounding case about a consumer, a sync, or a load path, every M1 check passes, and the business tree fills with cases no one can demo. ⚠️ Passing a case because its prose is clean is the exact failure this box catches. ⚠️ Conversely, do NOT fail a case merely for containing a technical-sounding noun: if a user or QC can demo the outcome, it is business — M7 asks what the case is ABOUT, not which words it uses.

If ANY box fails → verdict is NEEDS WORK; list each violated mandate ID with its concrete section/line citation in the Action Items.

Readability Checklist (MUST ATTENTION evaluate)

Before approving, verify code is easy to read, easy to maintain, easy to understand:

• Schema visibility — If function computes a data structure (object, map, config), a comment should show output shape so readers don't trace the code
• Non-obvious data flows — If data transforms through multiple steps (A → B → C), a brief comment should explain the pipeline
• Self-documenting signatures — Function params should explain their role; flag unused params
• Magic values — Unexplained numbers/strings should be named constants or have inline rationale
• Naming clarity — Variables/functions should reveal intent without reading the implementation

Output Format (per --type)

Pick the template matching --type. All templates lead with a Status/Verdict and ### Required/### Recommended tallies; the pbi/story/spec-tests shapes add their own evidence sections (preserved verbatim from the former refine-review, story-review, tdd-spec-review skills).

--type=design (default shape)

## Artifact Review

**Artifact Type:** [PBI | Story | Design | Test Spec]
**Artifact:** [Reference/title]
**Date:** {date}
**Verdict:** READY | NEEDS WORK

### Checklist Results
- [pass] [Item] — [evidence]
- [fail] [Item] — [what's missing/wrong]

### Action Items (if NEEDS WORK)
1. [Specific actionable item]

--type=pbi

## PBI Review Result

**Status:** PASS | WARN | FAIL
**Artifact:** {pbi-path}

### Required ({X}/{Y})

- ✅/❌ Check description

### Recommended ({X}/{Y})

- ✅/⚠️ Check description

### Issues Found

- ❌ FAIL: {issue}
- ⚠️ WARN: {issue}

### Verdict

{PROCEED | REVISE_FIRST}

--type=story

## Story Review Result

**Status:** PASS | WARN | FAIL
**Stories reviewed:** {count}
**Source PBI:** {pbi-path}

### AC Coverage Matrix

| Acceptance Criterion | Covered By Story | Status |
| -------------------- | ---------------- | ------ |

### Required ({X}/{Y})

- ✅/❌ Check description

### Recommended ({X}/{Y})

- ✅/⚠️ Check description

### Missing Stories

- {Any PBI AC not covered}

### Dependency Issues

- {Circular deps, missing ordering}

### Verdict

{PROCEED | REVISE_FIRST}

--type=spec-tests

## Test Spec Review Result

**Status:** PASS | WARN | FAIL
**TCs reviewed:** {count}
**Coverage:** {X}% of stories, {Y}% of acceptance criteria

### Coverage Matrix

| Story/AC | TC IDs | Happy | Error | Edge |
| -------- | ------ | ----- | ----- | ---- |

### Required ({X}/{Y})

- ✅/❌ Check description

### Recommended ({X}/{Y})

- ✅/⚠️ Check description

### Missing Coverage

- {Stories/AC without TCs}

### Verdict

{PROCEED | REVISE_FIRST}

Validated Fix + Full Re-Review (MANDATORY when fixes are applied)

Protocol: SYNC:double-round-trip-review + SYNC:fresh-context-review + SYNC:review-protocol-injection (all inlined above in this file).

Do not spawn a fresh sub-agent just to re-review the same finding set before fixing it. If the artifact needs work, fix actionable findings first, then restart the full artifact review over the current artifact. When that restarted review uses a fresh general-purpose sub-agent, use the canonical Agent template from SYNC:review-protocol-injection above. Artifact reviews (PBI, story, design spec, test spec) are NOT code — use subagent_type: "general-purpose", not "code-reviewer". When constructing the Agent call prompt:

1. Copy the Agent call shape from the SYNC:review-protocol-injection template verbatim
2. Set subagent_type: "general-purpose"
3. Embed the full verbatim body of these SYNC blocks (inlined above in this skill file): SYNC:evidence-based-reasoning, SYNC:rationalization-prevention, SYNC:understand-code-first (omit code-specific protocols like SYNC:bug-detection, SYNC:design-patterns-quality, SYNC:fix-layer-accountability which are not applicable to artifact files)
4. Set the Task as "Run a full fresh artifact review over the current {artifact-type} after fixes were applied. Focus on: implicit assumptions, missing coverage of edge cases / error scenarios, unverified cross-references, completeness gaps only visible on second reading, whether acceptance criteria are truly testable and measurable, and regressions introduced by fixes."
5. Set Target Files as the explicit artifact file path(s)
6. Set report path as plans/reports/artifact-review-rerun{N}-{date}.md

After sub-agent returns:

1. Read the sub-agent's report
2. Integrate findings as ## Re-Review {N} Findings in the main report — DO NOT filter or override
3. If NEEDS WORK: fix actionable artifact findings, then restart the full artifact review from the beginning
4. Repeated blocker cap: if the same blocker repeats across 2 full invocations with no progress, escalate via AskUserQuestion
5. Final verdict must incorporate findings from ALL review passes that actually ran

IMPORTANT Task Planning Notes (MUST ATTENTION FOLLOW)

• Always plan and break work into many small todo tasks using TaskCreate
• Always add a final review todo task to verify work quality and identify fixes/enhancements

---

Bulk Multi-Artifact Sweeps

For bulk multi-artifact review (10+ artifacts at once), use /changes-review — its Systematic Review Protocol categorizes the set and fires parallel sub-agents.

---

AI Agent Integrity Gate (NON-NEGOTIABLE)

Completion ≠ Correctness. Before reporting ANY work done, prove it:

1. Grep every removed name. Extraction/rename/delete touched N files? Grep confirms 0 dangling refs across ALL file types.
2. Ask WHY before changing. Existing values are intentional until proven otherwise. No "fix" without traced rationale.
3. Verify ALL outputs. One build passing ≠ all builds passing. Check every affected stack.
4. Evaluate pattern fit. Copying nearby code? Verify preconditions match — same scope, lifetime, base class, constraints.
5. New artifact = wired artifact. Created something? Prove it's registered, imported, and reachable by all consumers.

[IMPORTANT] Use TaskCreate to break ALL work into small tasks BEFORE starting — including tasks for each file read. This prevents context loss from long files. For simple tasks, AI MUST ATTENTION ask user whether to skip.

Prerequisites: MUST ATTENTION READ before executing:

OOP & DRY Enforcement: MANDATORY IMPORTANT MUST ATTENTION — flag duplicated patterns that should be extracted to a base class, generic, or helper. Classes in the same group or suffix (ex *Entity, *Dto, *Service, etc...) MUST ATTENTION inherit a common base (even if empty now — enables future shared logic and child overrides). Verify project has code linting/analyzer configured for the stack.

<!-- SYNC:nested-task-creation -->

Nested Task Expansion Contract — For workflow-step invocation, the [Workflow] ... row is only a parent container; the child skill still creates visible phase tasks.

1. Call TaskList first. If a matching active parent workflow row exists, set nested=true and record parentTaskId; otherwise run standalone.
2. Create one task per declared phase before phase work. When nested, prefix subjects [N.M] $skill-name — phase.
3. When nested, link the parent with TaskUpdate(parentTaskId, addBlockedBy: [childIds]).
4. Orchestrators must pre-expand a child skill's phase list and link the workflow row before invoking that child skill or sub-agent.
5. Mark exactly one child in_progress before work and completed immediately after evidence is written.
6. Complete the parent only after all child tasks are completed or explicitly cancelled with reason.

Blocked until: TaskList done, child phases created, parent linked when nested, first child marked in_progress.

<!-- /SYNC:nested-task-creation --> <!-- SYNC:project-reference-docs-guide -->

Project Reference Docs Gate — Run after task-tracking bootstrap and before target/source file reads, grep, edits, or analysis. Project docs override generic framework assumptions.

1. Identify scope: file types, domain area, and operation.
2. Read docs/project-config.json first — the project's machine-readable map. It is the single source of truth for THIS repo (modules/paths, framework + search keywords, test/E2E/integration run-commands, design system, architecture rules, workflow patterns); ground exact paths, run-commands, and conventions on it before investigating, planning, or coding — never assume framework defaults (CLAUDE.md + reference docs are derived from it). If it — or the docs index, lessons.md, CLAUDE.md, AGENTS.md, or any required reference doc — is missing or stale, auto-run /project-init or the narrow route (/project-config, /docs-init, /scan-all, /scan --target=<key>, /claude-md-init) first; if Codex mirrors or AGENTS.md are stale, ask the user to run /sync-codex (never auto-run it).
3. Required docs by trigger: always docs/project-reference/lessons.md; doc lookup docs-index-reference.md; review code-review-rules.md; backend/CQRS/API backend-patterns-reference.md; domain/entity domain-entities-reference.md; frontend/UI frontend-patterns-reference.md; styles/design scss-styling-guide.md + design-system/design-system-canonical.md; integration tests integration-test-reference.md; E2E e2e-test-reference.md; feature docs/specs feature-spec-reference.md + spec-system-reference.md + spec-principles.md; behavior/public-contract/spec-test-code sync workflow-spec-test-code-cycle-reference.md; derived spec index/ERD/reimplementation guides spec-system-reference.md + source Feature Specs under docs/specs/; architecture/new area project-structure-reference.md.
4. Read every required doc, then before target work state: Reference docs read: ... | Not applicable: ....

Ready when: scope evaluated, docs/project-config.json consulted, required docs checked/read or setup route completed, lessons.md confirmed, citation emitted.

<!-- /SYNC:project-reference-docs-guide --> <!-- SYNC:task-tracking-external-report -->

Task Tracking & External Report Persistence — Bootstrap this before execution; then run project-reference doc prefetch before target/source work.

1. Create a small task breakdown before target file reads, grep, edits, or analysis. On context loss, inspect the current task list first.
2. Mark one task in_progress before work and completed immediately after evidence; never batch transitions.
3. For plan/review work, create plans/reports/{skill}-{YYMMDD}-{HHmm}-{slug}.md before first finding.
4. Append findings after each file/section/decision and synthesize from the report file at the end.
5. Final output cites Full report: plans/reports/{filename}.

Blocked until: task breakdown exists, report path declared for plan/review work, first finding persisted before the next finding.

<!-- /SYNC:task-tracking-external-report --> <!-- SYNC:critical-thinking-mindset -->

Critical Thinking Mindset — Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence >80% to act. Anti-hallucination: Never present guess as fact — cite sources for every claim, admit uncertainty freely, self-check output for errors, cross-reference independently, stay skeptical of own confidence — certainty without evidence root of all hallucination.

<!-- /SYNC:critical-thinking-mindset --> <!-- SYNC:evidence-based-reasoning -->

Evidence-Based Reasoning — Speculation is FORBIDDEN. Every claim needs proof.

1. Cite file:line, grep results, or framework docs for EVERY claim
2. Declare confidence: >80% act freely, 60-80% verify first, <60% DO NOT recommend
3. Cross-service validation required for architectural changes
4. "I don't have enough evidence" is valid and expected output

BLOCKED until: - [ ] Evidence file path (file:line) - [ ] Grep search performed - [ ] 3+ similar patterns found - [ ] Confidence level stated

Forbidden without proof: "obviously", "I think", "should be", "probably", "this is because" If incomplete → output: "Insufficient evidence. Verified: [...]. Not verified: [...]."

<!-- /SYNC:evidence-based-reasoning --> <!-- SYNC:understand-code-first -->

Understand Code First — HARD-GATE: Do NOT write, plan, or fix until you READ existing code.

1. Search 3+ similar patterns (grep/glob) — cite file:line evidence
2. Read existing files in target area — understand structure, base classes, conventions
3. Run python .claude/scripts/code_graph trace <file> --direction both --json when .code-graph/graph.db exists
4. Map dependencies via connections or callers_of — know what depends on your target
5. Write investigation to .ai/workspace/analysis/ for non-trivial tasks (3+ files)
6. Re-read analysis file before implementing — never work from memory alone. — why: long context drifts from the file; the file is ground truth
7. NEVER invent new patterns when existing ones work — match exactly or document deviation. — why: divergent patterns fragment the codebase and slow every future reader

BLOCKED until: - [ ] Read target files - [ ] Grep 3+ patterns - [ ] Graph trace (if graph.db exists) - [ ] Assumptions verified with evidence

<!-- /SYNC:understand-code-first --> <!-- SYNC:double-round-trip-review -->

Validated-Finding Fix + Full Re-Review Loop — Re-review is triggered by a validated finding fix cycle, not by a round number. Review purpose: review → validate findings → fix validated findings → full re-review until a complete review pass clears the round's exit bar (see Severity floor below). A clean review ENDS the loop — no further rounds required.

aka Self-Review Convergence Loop. The name is historical — there is NO 2-round cap; "double-round-trip" only means a validated-finding fix cycle forces at least one fresh re-review. It runs until a clean pass, bounded by the 3-round ceiling below.

Round cap — 3 rounds MAX (a ceiling, NEVER a target). A clean pass ENDS the loop immediately at ANY round — round 1 included; the cap never obliges you to keep spinning. Hitting round 3 with blocking findings still open (severity floor applied) → STOP and escalate via AskUserQuestion with the still-open findings listed; NEVER emit a silent "good enough" PASS on cap exhaustion, and NEVER let the cap substitute for the clean-review requirement. The 2-repeated-no-progress blocker rule stays an EARLIER exit — escalate at whichever trips first.

Severity floor — from round 3, LOW stops blocking. The exit bar tightens by round, so the loop converges on consequence instead of spinning on polish:

Define one predicate everywhere: blocking_findings(round, findings) returns all validated findings in rounds 1–2 and only validated CRITICAL/HIGH/MEDIUM findings in round 3+. A binary gate (test-green, security must-fix, required artifact) is exempt only when its owning invariant explicitly says so.

Round	Exit bar — loop ENDS when the fresh full review has…	Must be fixed to continue
1-2	zero validated findings at ANY severity	CRITICAL · HIGH · MEDIUM · LOW
3+	zero validated CRITICAL / HIGH / MEDIUM findings — LOW-only is a PASS	CRITICAL · HIGH · MEDIUM only

From round 3 onward LOW findings are NOT required to be fixed: a round whose validated findings are ALL LOW ENDS the loop immediately — do not open another round for them. Severity tiers are SYNC:severity-rubric (CRITICAL block-merge · HIGH must-fix · MEDIUM should-fix · LOW nice-to-fix); rounds 1-2 are unchanged, so an easy LOW still gets fixed early when it is cheap.

Severity-floor rules:

• Never silently drop a deferred LOW. Every unfixed LOW is listed in the final report under ## Deferred LOW Findings (severity floor, round ≥3) with file, line, and description, so the owner can schedule it. Dropping it from the report is a protocol violation, not a clean pass.
• Never re-tier a finding to trigger the exit. Downgrading a real CRITICAL/HIGH/MEDIUM to LOW so the loop can end is a FALSE PASS. Severity is set by consequence per SYNC:severity-rubric before the round bar is applied — never after, and never with the exit in view. — why: a floor that can be reached by relabeling is not a floor.
• The floor bounds the loop, not the standard. It ends iteration; it never authorizes shipping a known CRITICAL/HIGH/MEDIUM, and it never lowers the finding-survival bar that admits a finding in the first place.
• The floor never applies to a hard gate. Test-green gates (a suite must actually pass), security must-fix gates, and any gate whose criterion is binary rather than severity-rated are unaffected — a failing test is a failure, not a LOW finding.

Universal scope (any new output/judgment): any newly produced output or judgment gets ≥1 self-review; any new judgment gets ≥1 /why-review --validate-findings pass; anything flagged to re-check is re-checked ≥1 time — before that output is treated as final. This loop is the default convergence contract for ANY work-producing skill, not review skills only.

Routing invariant (author-facing): a skill that validates findings MUST route them through /why-review --validate-findings (the terminal validator) — NEVER fork an inline finding-validation. Routing through why-review is what makes the finding-survival bar and this loop apply; the verify-review-validate-coverage sensor enforces this exact route mechanically.

Round 1: Main-session review. Read target files, build understanding, note issues. Output findings + verdict (PASS / FAIL).

Decision after Round 1:

• No issues found (PASS, zero findings) → review ENDS. Do NOT spawn a fresh sub-agent for confirmation.
• blocking_findings(round, findings) is non-empty → run the active review skill's findings-validation gate first; for review skills the default gate is /why-review --validate-findings <report-path>. Fix only validated findings, then restart the full review protocol from the beginning with a fresh task breakdown.

Fresh full re-review after every fix cycle: Re-run the whole review protocol over the current full target. When sub-agents are part of that protocol, spawn NEW Agent calls — never reuse prior agents. Reviewers re-read ALL files from scratch with ZERO memory of prior rounds. See SYNC:fresh-context-review for the spawn mechanism and SYNC:review-protocol-injection for the canonical Agent prompt template. Each fresh full review must catch:

• Cross-cutting concerns missed in the prior round
• Interaction bugs between changed files
• Convention drift (new code vs existing patterns)
• Missing pieces that should exist but don't
• Subtle edge cases the prior round rationalized away
• Regressions introduced by the fixes themselves

Loop termination: After each full re-review, repeat the same decision against that round's exit bar: bar cleared → END; blocking findings remain → validate findings → fix → restart from the first review phase. Rounds 1-2 clear on zero findings at any severity; from round 3 the bar is zero CRITICAL/HIGH/MEDIUM, so a LOW-only round ENDS the loop (deferred LOWs go in the report). Capped at 3 rounds. Escalate via AskUserQuestion at whichever comes first: the same validated finding repeats for 2 full invocations with no progress · a fix requires product/owner input · round 3 completes with CRITICAL/HIGH/MEDIUM still open. NEVER loop past 3 rounds, and NEVER convert cap exhaustion into a PASS.

Rules:

• A clean Round 1 ENDS the review — no mandatory Round 2
• From round 3 on, a round whose validated findings are ALL LOW ENDS the loop — never open round N+1 to fix LOW alone; list those LOWs as deferred instead
• NEVER re-tier a CRITICAL/HIGH/MEDIUM down to LOW to reach the round-3 exit — severity is assigned by consequence before the bar is applied
• NEVER fix unvalidated findings; validate first using the caller's validation gate
• Every surviving finding must additionally clear the finding-survival bar defined in why-review's Findings Validation Routine (a deliberately higher bar than the generic act-gate — "keep this finding?" is a stricter question than "act on this evidence?"); a finding below the bar is demoted or dropped, not kept
• NEVER skip the full re-review after a fix cycle (every fix invalidates the prior verdict)
• NEVER reuse a sub-agent across rounds — every iteration that uses sub-agents spawns NEW Agent calls
• Main agent READS sub-agent reports but MUST NOT filter, reinterpret, or override findings
• The 3-round cap NEVER replaces the clean-review requirement — it bounds runaway looping, it does not authorize shipping an un-clean review; a clean pass ends the loop early at any round, and cap exhaustion escalates rather than passes
• Enforce the round cap of 3 alongside the 2 repeated-no-progress blocker rule; both are escalation triggers, neither is a completion criterion
• Track recursive invocation count and repeated blockers in conversation context (session-scoped)
• Final verdict must incorporate ALL rounds executed

Report must include ## Round N Findings (Fresh Sub-Agent) for every round N≥2 that was executed, plus ## Deferred LOW Findings (severity floor, round ≥3) whenever the loop ended on the round-3+ bar with LOWs still open.

<!-- /SYNC:double-round-trip-review --> <!-- SYNC:fresh-context-review -->

Fresh Context Re-Review — Eliminate orchestrator confirmation bias after fixes by restarting the full review with isolated sub-agents where applicable.

Why: The main agent knows what it (or /feature-implement) just fixed and rationalizes findings accordingly. A fresh sub-agent has ZERO memory, re-reads from scratch, and catches what the main agent dismissed. Sub-agent bias is mitigated by (1) fresh context, (2) verbatim protocol injection, (3) main agent not filtering the report.

When: ONLY after a validated-finding fix cycle. A review round that finds zero issues ENDS the loop — do NOT spawn a confirmation sub-agent. A review round that finds issues triggers: validate findings → fix → full review restart from the first phase.

How:

1. Start a NEW full review invocation/task breakdown; when that protocol calls for agents, spawn NEW Agent tool calls — use code-reviewer subagent_type for code reviews, general-purpose for plan/doc/artifact reviews
2. Inject ALL required review protocols VERBATIM into the prompt — see SYNC:review-protocol-injection for the full list and template. Never reference protocols by file path; AI compliance drops behind file-read indirection (see SYNC:shared-protocol-duplication-policy)
3. Sub-agent re-reads ALL target files from scratch via its own tool calls — never pass file contents inline in the prompt
4. Sub-agent writes structured report to plans/reports/{review-type}-round{N}-{date}.md
5. Main agent reads the report, integrates findings into its own report, DOES NOT override or filter

Rules:

• SKIP fresh sub-agent when the prior full review found zero issues (no fixes = nothing new to verify)
• NEVER skip the full review restart after a fix cycle — every fix invalidates the prior verdict
• NEVER reuse a sub-agent across rounds — every fresh round spawns a NEW Agent call
• Continue until a complete full review pass clears that round's exit bar per SYNC:double-round-trip-review: rounds 1-2 → zero findings at any severity; round 3+ → zero CRITICAL/HIGH/MEDIUM, so a round whose validated findings are ALL LOW ENDS the loop (list those LOWs as deferred instead of spawning another round). If the same blocker repeats 3 times with no progress, escalate via AskUserQuestion
• Track iteration count and repeated blockers in conversation context (session-scoped, no persistent files)

<!-- /SYNC:fresh-context-review --> <!-- SYNC:review-protocol-injection -->

Review Protocol Injection — Every fresh sub-agent review prompt MUST embed 11 protocol blocks VERBATIM. The template below has ALL 11 bodies already expanded inline. Copy the template wholesale into the Agent call's prompt field at runtime, replacing only the {placeholders} in Task / Round / Reference Docs / Target Files / Output sections with context-specific values. Do NOT touch the embedded protocol sections.

Why inline expansion: Placeholder markers would force file-read indirection at runtime. AI compliance drops significantly behind indirection (see SYNC:shared-protocol-duplication-policy). Therefore the template carries all 11 protocol bodies pre-embedded.

Subagent Type Selection

• code-reviewer — for code reviews (reviewing source files, git diffs, implementation)
• general-purpose — for plan / doc / artifact reviews (reviewing markdown plans, docs, specs)

Canonical Agent Call Template (Copy Verbatim)

Agent({
  description: "Fresh Round {N} review",
  subagent_type: "code-reviewer",
  prompt: `
## Task
{review-specific task — e.g., "Review all uncommitted changes for code quality" | "Review plan files under {plan-dir}" | "Review integration tests in {path}"}

## Round
Round {N}. You have ZERO memory of prior rounds. Re-read all target files from scratch via your own tool calls. Do NOT trust anything from the main agent beyond this prompt.

## Protocols (follow VERBATIM — these are non-negotiable)

### Spec ↔ Tests ↔ Code Triangulation
DO THIS FIRST — before any per-protocol check below. The review target is the WHOLE PACKAGE, not the diff alone: load the behavior's spec (§3 ACs / §4 BRs / §8 TCs), its tests, and the changed code TOGETHER, and reason about their mutual consistency BEFORE judging any one in isolation.
1. Locate all three faces: the Feature Spec section(s) governing the changed behavior, the tests that guard it, and the production code that implements it. A missing face is itself a finding (SPEC-GAP / TEST-GAP / DEAD-SPEC).
2. Triangulate pairwise — every disagreement is a finding; classify which face is wrong:
   - code vs spec: behavior the code does that no §3/§4/§8 rule describes → CODE-EXTRA or SPEC-STALE; a [HARD] §4 rule or §5 invariant with no enforcing code path → CODE-WRONG.
   - tests vs spec: a §8 TC with no test, or a test asserting behavior no TC/rule names → TEST-GAP or SPEC-SILENT.
   - tests vs code: a changed code path with no covering test → TEST-GAP; a test that still passes against a deliberately broken invariant → WEAK-TEST (apply the mutation thinking in Bug Detection).
3. Hidden-rule capture: any invariant the code enforces but the spec never states (SPEC-SILENT) MUST be surfaced as a finding to add into §3/§4/§8 AND guarded with a test — the enrichment loop, never a silent pass.
4. Only after the three faces agree — or every disagreement is logged as a finding — proceed to the per-protocol checks below; when enrichment adds spec/test content, re-review the package against the enriched spec.
NEVER mark review PASS while any spec/test/code face disagrees without a logged finding. The diff is the entry point; the package is the unit of judgment.

### Evidence-Based Reasoning
Speculation is FORBIDDEN. Every claim needs proof.
1. Cite file:line, grep results, or framework docs for EVERY claim
2. Declare confidence: >80% act freely, 60-80% verify first, <60% DO NOT recommend
3. Cross-service validation required for architectural changes
4. "I don't have enough evidence" is valid and expected output
BLOCKED until: Evidence file path (file:line) provided; Grep search performed; 3+ similar patterns found; Confidence level stated.
Forbidden without proof: "obviously", "I think", "should be", "probably", "this is because".
If incomplete → output: "Insufficient evidence. Verified: [...]. Not verified: [...]."

### Bug Detection
MUST check categories 1-4 for EVERY review. Never skip.
1. Null Safety: Can params/returns be null? Are they guarded? Optional chaining gaps? .find() returns checked?
2. Boundary Conditions: Off-by-one (< vs <=)? Empty collections handled? Zero/negative values? Max limits?
3. Error Handling: Try-catch scope correct? Silent swallowed exceptions? Error types specific? Cleanup in finally?
4. Resource Management: Connections/streams closed? Subscriptions unsubscribed on destroy? Timers cleared? Memory bounded?
5. Concurrency (if async): Missing await? Race conditions on shared state? Stale closures? Retry storms?
6. Stack-Specific: Check the configured language/runtime pitfalls and framework-specific failure modes discovered from local code.
Classify: CRITICAL (crash/corrupt) → FAIL | HIGH (incorrect behavior) → FAIL | MEDIUM (edge case) → WARN | LOW (defensive) → INFO.

### Design Patterns Quality
Priority checks for every code change:
1. DRY via OOP: Same-suffix classes (*Entity, *Dto, *Service) MUST share base class. 3+ similar patterns → extract to shared abstraction.
2. Right Responsibility: Logic in LOWEST layer (Entity > Domain Service > Application Service > Controller). Never business logic in controllers.
3. SOLID: Single responsibility (one reason to change). Open-closed (extend, don't modify). Liskov (subtypes substitutable). Interface segregation (small interfaces). Dependency inversion (depend on abstractions).
4. After extraction/move/rename: Grep ENTIRE scope for dangling references. Zero tolerance.
5. YAGNI gate: NEVER recommend patterns unless 3+ occurrences exist. Don't extract for hypothetical future use.
Anti-patterns to flag: God Object, Copy-Paste inheritance, Circular Dependency, Leaky Abstraction.

### Logic & Intention Review
Verify WHAT code does matches WHY it was changed.
1. Change Intention Check: Every changed file MUST serve the stated purpose. Flag unrelated changes as scope creep.
2. Happy Path Trace: Walk through one complete success scenario through changed code.
3. Error Path Trace: Walk through one failure/edge case scenario through changed code.
4. Acceptance Mapping: If plan context available, map every acceptance criterion to a code change.
5. Tests Verify Intent: For test/spec changes, verify tests name the protected business rule or invariant and would fail if that intent breaks.
6. Migration Test Exclusion: Do not write tests for migration code. Schema/data migrations are one-time execution paths, not core application logic.
NEVER mark review PASS without completing both traces (happy + error path).

### Test Spec Verification
Map changed code to test specifications.
1. Identify the project's test/spec format from existing docs, test-case files, BDD feature files, or spec folders.
2. Every changed code path MUST map to a corresponding test case/spec (or flag as "needs test case").
3. New functions/endpoints/handlers → flag for test spec creation.
4. Migration files are excluded from test/spec creation; schema/data migrations are one-time execution paths, not core application logic.
5. If spec evidence fields exist, verify they point to actual code (file:line, not stale references).
6. Verify each meaningful test case names the business intent/invariant; flag behavior-only cases that only mirror implementation details.
7. Auth/data changes → verify corresponding authorization and data-state test cases exist.
8. If no specs exist for a changed path → log the gap and recommend the project's test-spec workflow.
NEVER skip test mapping. Untested code paths are the #1 source of production bugs.

### Behavioral Delta Matrix
MANDATORY for any bugfix review. Produce input-state × pre-fix × post-fix × delta table BEFORE writing verdict.
- Minimum 3 rows; include at least one row OUTSIDE the original bug report.
- Any "REGRESSION" delta → review returns FAIL until a preservation test is added.
- Narrative descriptions do NOT substitute for the matrix.
Example rows (external-record sync fix):
| Input                 | Pre-fix | Post-fix                  | Delta      |
| --------------------- | ------- | ------------------------- | ---------- |
| Record exists (valid) | Reused  | Always recreated → orphan | REGRESSION |
| Record missing (404)  | Error   | Recreated                 | Fixed      |

### Fix-Layer Accountability
NEVER fix at the crash site. Trace the full flow, fix at the owning layer. The crash site is a SYMPTOM, not the cause.
MANDATORY before ANY fix:
1. Trace full data flow — Map the complete path from data origin to crash site across ALL layers (storage → backend → API → frontend → UI). Identify where bad state ENTERS, not where it CRASHES.
2. Identify the invariant owner — Which layer's contract guarantees this value is valid? Fix at the LOWEST layer that owns the invariant, not the highest layer that consumes it.
3. One fix, maximum protection — If fix requires touching 3+ files with defensive checks, you are at the wrong layer — go lower.
4. Verify no bypass paths — Confirm all data flows through the fix point. Check for direct construction skipping factories, clone/spread without re-validation, raw data not wrapped in domain models, mutations outside the model layer.
BLOCKED until: Full data flow traced (origin → crash); Invariant owner identified with file:line evidence; All access sites audited (grep count); Fix layer justified (lowest layer that protects most consumers).
Anti-patterns (REJECT): "Fix it where it crashes" (crash site ≠ cause site, trace upstream); "Add defensive checks at every consumer" (scattered defense = wrong layer); "Both fix is safer" (pick ONE authoritative layer).

### Rationalization Prevention
AI skips steps via these evasions. Recognize and reject:
- "Too simple for a plan" → Simple + wrong assumptions = wasted time. Plan anyway.
- "I'll test after" → RED before GREEN. Write/verify test first.
- "Already searched" → Show grep evidence with file:line. No proof = no search.
- "Just do it" → Still need TaskCreate. Skip depth, never skip tracking.
- "Just a small fix" → Small fix in wrong location cascades. Verify file:line first.
- "Code is self-explanatory" → Future readers need evidence trail. Document anyway.
- "Combine steps to save time" → Combined steps dilute focus. Each step has distinct purpose.

### Graph-Assisted Investigation
MANDATORY when .code-graph/graph.db exists.
HARD-GATE: MUST run at least ONE graph command on key files before concluding any investigation.
Pattern: Grep finds files → trace --direction both reveals full system flow → Grep verifies details.
- Investigation/Scout: trace --direction both on 2-3 entry files
- Fix/Debug: callers_of on buggy function + tests_for
- Feature/Enhancement: connections on files to be modified
- Code Review: tests_for on changed functions
- Blast Radius: trace --direction downstream
CLI: python .claude/scripts/code_graph {command} --json. Use --node-mode file first (10-30x less noise), then --node-mode function for detail.

### Understand Code First
HARD-GATE: Do NOT write, plan, or fix until you READ existing code.
1. Search 3+ similar patterns (grep/glob) — cite file:line evidence.
2. Read existing files in target area — understand structure, base classes, conventions.
3. Run python .claude/scripts/code_graph trace <file> --direction both --json when .code-graph/graph.db exists.
4. Map dependencies via connections or callers_of — know what depends on your target.
5. Write investigation to .ai/workspace/analysis/ for non-trivial tasks (3+ files).
6. Re-read analysis file before implementing — never work from memory alone.
7. NEVER invent new patterns when existing ones work — match exactly or document deviation.
BLOCKED until: Read target files; Grep 3+ patterns; Graph trace (if graph.db exists); Assumptions verified with evidence.

## Reference Docs (READ before reviewing)
- `.claude/docs/development-rules.md` — canonical development rules, code-quality guidelines, and pre-commit checklist
- docs/project-reference/code-review-rules.md
- {skill-specific reference docs — e.g., integration-test-reference.md for integration-test-review; backend-patterns-reference.md for backend reviews; frontend-patterns-reference.md for frontend reviews}

## Target Files
{explicit file list OR "run git diff to see uncommitted changes" OR "read all files under {plan-dir}"}

## Output
Write a structured report to plans/reports/{review-type}-round{N}-{date}.md with sections:
- Status: PASS | FAIL
- Issue Count: {number}
- Critical Issues (with file:line evidence)
- High Priority Issues (with file:line evidence)
- Medium / Low Issues
- Cross-cutting findings

Return the report path and status to the main agent.
Every finding MUST have file:line evidence. Speculation is forbidden.
`
})

Rules

• DO copy the template wholesale — including all 11 embedded protocol sections
• DO replace only the {placeholders} in Task / Round / Reference Docs / Target Files / Output sections with context-specific content
• DO choose code-reviewer subagent_type for code reviews and general-purpose for plan / doc / artifact reviews
• DO NOT paraphrase, summarize, or skip any protocol section
• DO NOT pass file contents inline — the sub-agent reads via its own tool calls so it has a fresh context
• DO NOT reference protocols by file path or tag name — the bodies are already embedded above
• DO NOT introduce placeholder markers for the protocols — they must stay literally expanded

<!-- /SYNC:review-protocol-injection --> <!-- SYNC:ai-mistake-prevention -->

AI Mistake Prevention — Failure modes to avoid on every task:

Re-read files after context changes. Context compaction, resume, or long-running work can make memory stale; verify current files before acting. Verify generated content against source evidence. AI hallucinates APIs, names, claims, and document facts. Check the relevant source before documenting or referencing. Check downstream references before deleting or renaming. Removing an artifact can stale docs, generated mirrors, configs, and callers; map references first. Trace the full impact chain after edits. Changing a definition can miss derived outputs and consumers. Follow the affected chain before declaring done. Verify ALL affected outputs, not just the first. One green check is not all green checks; validate every output surface the change can affect. Assume existing values are intentional — ask WHY before changing OR flagging one as a defect. Before changing or reporting a constant, limit, flag, cutoff, wording, or pattern, read nearby context and history, the CALLER's ordering, and 2+ sibling call sites of the same convention. A doc stating WHAT without WHY is missing rationale, not proof of a missing guard. Surface ambiguity before acting — don't pick silently. Multiple valid interpretations require an explicit question or stated assumption with risk. Assert the outcome your system owns, not the intermediate state your infrastructure owns. When verifying async work, assert the final business state — never the delivery/retry bookkeeping held in shared infrastructure that any co-running process can write. Such a check passes when run alone and flakes the moment anything else shares that infrastructure. Keep shared guidance role-relevant. Universal guidance must help every receiving skill or agent; code-specific obligations belong only in code-specific protocols.

<!-- /SYNC:ai-mistake-prevention --> <!-- SYNC:ui-intent-layer -->

[BLOCKING] Capture a tech-agnostic UI/UX intent layer in every UI-bearing spec — a reader must be able to visualize how the feature works without naming any technology. When the feature has a user interface, the spec MUST ATTENTION carry an interaction-surface section so the application — not just its API — can be rebuilt on any stack:

1. View Inventory — list each view/screen by its UX ROLE and purpose (e.g. "list of items", "item editor", "confirmation step") and what information it presents. Describe by role, never by an implementation name.
2. Navigation Map — how a user moves between views: entry points, transitions, and exits. Trace how this surface connects to neighboring features already in the system.
3. Key observable UI States — the distinct states a user can observe per view (empty, loading, populated, error, success, permission-denied, etc.) — described as what the user perceives, not how it is rendered.
4. Per-story interaction flow — for each user story, the step-by-step click/action path from intent to outcome, cross-referenced to the logical IDs the spec already owns (US-/OP-/BR-).
5. Couple to the companion design artifact — keep deep visual fidelity (layout, tokens, pixel detail) OUT of the spec; it lives in the linked design-spec/mockup. Record that companion's path in the spec frontmatter so the spec stays the navigable hub.

M1-clean (NON-NEGOTIABLE): the prose names ZERO frameworks, routes/URLs, CSS, or component-class names — only roles, information, states, and flows. Technology detail belongs in the companion design artifact, never here.

Skip ONLY when the feature is backend-only (no UI) — state that reason explicitly in the section.

<!-- /SYNC:ui-intent-layer --> <!-- SYNC:severity-rubric -->

Severity Rubric — Classify every finding by consequence, not by how easy it is to fix. One scale across all reviews so a "High" means the same thing everywhere.

Severity	Action	Definition
CRITICAL	Block merge	Silent runtime failure, data corruption, validation bypass, security hole
HIGH	Must fix	Incorrect behavior, invariant gap, architectural violation
MEDIUM	Should fix	Design debt, maintainability, likely future bug
LOW	Nice to fix	Convention, documentation, minor clarity

Score-based skills map their numeric scale onto these tiers — do not invent a parallel vocabulary:

• 0-2 criterion scoring (e.g. production-readiness-review): 0 = CRITICAL/HIGH (criterion unmet, blocks production readiness), 1 = MEDIUM (partial, should fix), 2 = pass (no finding).
• Two-axis scoring (e.g. performance-review, impact × likelihood): map the resulting cell to the nearest tier — high-impact + high-likelihood → CRITICAL/HIGH; low-impact OR low-likelihood → MEDIUM/LOW.

A finding's tier drives the gate: CRITICAL/HIGH must be resolved or explicitly accepted by the owner before PASS; MEDIUM/LOW may ship with a tracked follow-up.

<!-- /SYNC:severity-rubric --> <!-- SYNC:goal-contract-satisfaction-loop -->

Goal Contract Satisfaction Loop — Persist the user goal in an external file, execute against it, and loop review/fix until every saved required criterion passes or a blocker escalates. Bounded closed loop — NEVER open-ended autonomous exploration.

1. Resolve the active goal (in order): active plan goal.md → plans/goals/{YYMMDD-HHmm}-{slug}/goal.md → create a new Goal Contract from the current user request (template: .claude/templates/goal-contract-template.md).
2. Required sections: Original Request, Purpose, Success Criteria (checkboxes; mark required vs optional), Constraints, Evidence Required, Iteration Log, Goal Satisfaction matrix.
3. Before work: read the active goal and map planned work to saved success criteria — execution serves the saved criteria, never chat memory alone.
4. After execution/verification: append an Iteration Log entry — result, evidence references (file:line, command output, report path), remaining gaps.
5. Review gate: emit a Goal Satisfaction matrix — | Success Criterion | Evidence | Status | with PASS/FAIL/BLOCKED. Overall PASS requires every required criterion PASS.
6. Loop rule (retry): required criterion FAIL → validate the gap is real → fix → re-review only the affected criteria. Stop cleanly when all required criteria PASS.
7. Escalation rule (stop): two consecutive iterations with no criterion progressing, or a blocker needing user input → mark the criterion BLOCKED with a user-facing reason and escalate. NEVER loop indefinitely.
8. Skip rule: tiny conversational tasks may skip the goal file ONLY with a recorded one-line reason. User-accepted gate skips are recorded in the goal file with reason and scope.
9. Security: NEVER store secrets, tokens, credentials, or private customer data in goal files — store evidence references and redact sensitive values.

Blocked until: active goal resolved (or skip reason recorded) · saved success criteria read before edits · iteration evidence appended after execution · Goal Satisfaction matrix emitted before any PASS verdict.

<!-- /SYNC:goal-contract-satisfaction-loop --> <!-- SYNC:trade-off-interrogation-gate -->

Trade-Off Interrogation Gate — ALWAYS ask these THREE questions before ANY verdict, score, finding, or recommendation — about the thing under review AND about every recommendation YOU make. — why: naming a benefit without its price is an endorsement, not a review; the costliest trade-offs are the ones nobody wrote down.

1. Is there any trade-off? Name what it SACRIFICES. "None" / "pure win" is an unfinished analysis, NOT an answer — to claim none, state which dimensions you checked and why each is unaffected: future change cost · complexity · performance/latency · memory/cost · coupling · reversibility · migration burden · operational load · blast radius · security posture · testability · team skill/ramp · delivery time · UX.
2. Is it worth it? Weigh gain against sacrifice EXPLICITLY — what is gained (with a metric) · what it costs · WHO pays · WHEN it comes due — then emit WORTH IT / NOT WORTH IT / UNCLEAR. "Better" with no metric and no cost FAILS this question. NOT WORTH IT → withdraw or replace the recommendation, never keep it as-is.
3. Is the trade-off material enough to CONFIRM WITH THE USER? A material trade-off is the user's call, never yours. MATERIAL when ANY holds: irreversible / one-way door (data migration, public contract, storage format, vendor lock-in) · cost shifted onto someone else (another team, ops/on-call, future maintainer, end user) · one quality attribute traded for another (correctness↔speed, security↔convenience, latency↔cost, simplicity↔flexibility) · a boundary crossed (client↔server tier, service contract, event contract, shared library) · a high-consequence path (auth, money, data integrity, breaking change, High/Medium residual risk) · the worth-it verdict is UNCLEAR.

MATERIAL → STOP and confirm via AskUserQuestion BEFORE the verdict stands — state the trade-off, both options, what each sacrifices, and your recommendation. NOT material → record it inline with a one-line justification and proceed.

Non-asking execution contexts — ESCALATE BY HANDOFF, never by silence. AskUserQuestion reaches only the main interactive agent: a sub-agent cannot ask the user, and a terminal/verdict-only mode asks nothing by design. When you are running in such a context, the obligation is redirected, never waived — do ALL of: (a) complete questions 1 and 2 normally; (b) decide materiality and record it in the Trade-Off Assessment row with confirmed? = NO — cannot ask from this context; (c) name the unconfirmed MATERIAL trade-off explicitly in your returned summary/verdict so the CALLER (or parent orchestrator) escalates it via AskUserQuestion on your behalf — a material trade-off mentioned only inside a report file on disk is NOT a handoff; (d) do not emit an unqualified PASS — mark the verdict as carrying an unconfirmed material trade-off, so the caller's gate stays closed until the user answers. The caller inherits the escalation duty the moment it reads your return.

This carve-out is about reachability, not convenience: it applies ONLY where the tool genuinely cannot reach the user (spawned sub-agent, terminal validate/verdict-only mode, non-interactive/headless run). It is NEVER a licence to skip the question, to self-approve a one-way door, or to downgrade materiality because asking is inconvenient — if you CAN ask, you MUST ask.

Emit a Trade-Off Assessment row per reviewed decision and per recommendation: | decision | sacrifices | gain (metric) | who pays, when | WORTH IT/NOT/UNCLEAR | material? | confirmed? |.

BLOCKED until: trade-off named (or dimensions-checked justification given) · worth-it verdict emitted · materiality decided · every MATERIAL trade-off either confirmed with the user OR — in a non-asking context — handed off in the returned verdict for the caller to confirm. A MATERIAL trade-off that is neither confirmed nor handed off can NEVER be PASS, and NEVER gets buried as a Low-severity note.

NEVER answer "no trade-off" without checking · decide a material trade-off silently on the user's behalf · let convergence/delivery pressure authorize walking through a one-way door · bundle several material trade-offs into one vague "proceed?".

<!-- /SYNC:trade-off-interrogation-gate --> <!-- SYNC:ui-ux-design-principles -->

UI/UX Design Principles (Rev 1.0 — 40 clauses, web + mobile) — the working rule set for ANY task that designs, plans, implements, or reviews a user interface. Applies to BOTH platforms unless a clause names one (§8 is mobile/touch). Cite clauses by ID: UI-3.1, UI-8.2.

Precedence: project design-system / SCSS / frontend-pattern docs OUTRANK these clauses; these clauses outrank generic taste. A genuine conflict is SURFACED to the user with both sides — NEVER resolved silently. — why: the project's own recorded decision is the authority; these clauses are the default when it is silent.

1.0 Visual Hierarchy & Layout

• UI-1.1 One focal point per screen. Two elements competing for first read → demote one.
• UI-1.2 Signal order: size → weight → colour → position. Use the cheapest signal that works before adding another.
• UI-1.3 Group by proximity, NEVER by border. Whitespace separates cleanly; boxes inside boxes do not.
• UI-1.4 Align to a shared edge. Every unexplained indent reads as an accident.
• UI-1.5 Design the empty, loading and error state FIRST. The full state is the easy one.

2.0 Typography

• UI-2.1 Max 2 families, 3 weights each. More variety reads as inconsistency, not range.
• UI-2.2 Body text 16px web, 17px mobile. NEVER below 14px for anything a user must read.
• UI-2.3 Line length 45–75 characters. Constrain the measure, not the container.
• UI-2.4 Leading scales inversely with size: 1.5 body, 1.1–1.2 display.
• UI-2.5 Fixed type scale — 6 named steps shared with engineering. NEVER one-off sizes.

3.0 Colour & Contrast

• UI-3.1 Contrast 4.5:1 text, 3:1 UI edges. Measure it — NEVER judge by eye on a bright screen.
• UI-3.2 One accent, one job. An accent that is everywhere points at nothing.
• UI-3.3 Colour NEVER carries meaning alone — pair it with an icon, label or position.
• UI-3.4 Dark mode is NOT inverted light mode. Lift surfaces to signal elevation; soften pure-white text.

4.0 Spacing & Grid

• UI-4.1 One spacing unit, multiplied — 4px or 8px base; every gap a multiple of it.
• UI-4.2 Space belongs to the container, not the child. Use gap; reserve margins for exceptions.
• UI-4.3 Tighter inside, looser between — inner padding always smaller than the gap to the next group.
• UI-4.4 Breakpoints follow content, not devices. Break where the layout stops working.

5.0 Interaction & Feedback

• UI-5.1 Every action gets a response under 100ms, even when the result takes longer.
• UI-5.2 Specify all 5 states — default, hover, focus, active, disabled — plus loading where it applies.
• UI-5.3 Prefer undo over confirmation. Confirm ONLY what cannot be reversed.
• UI-5.4 Motion clarifies cause and effect: 150–250ms, ease-out, honours reduced-motion.
• UI-5.5 Keep the visible focus ring. Restyle it if it clashes; NEVER remove it.

6.0 Navigation & IA

• UI-6.1 Every screen answers: where am I, what's here, where next.
• UI-6.2 Max 5 top-level destinations. Depth beats a crowded first level.
• UI-6.3 Label by the user's word, not the internal one. Team vocabulary is not a taxonomy.
• UI-6.4 Every state deserves a URL or a back path. Deep links and hardware back must land somewhere sensible.

7.0 Forms & Input

• UI-7.1 Ask for less. Every field needs a reason it exists today.
• UI-7.2 Labels stay visible. Placeholders are hints, NEVER labels.
• UI-7.3 Validate on blur, not on keystroke. Errors sit next to the field and say how to fix it.
• UI-7.4 Match keyboard to data type — correct input type, autocomplete and autocapitalise on every field.
• UI-7.5 NEVER lose entered data. Preserve input across errors, navigation and refresh.

8.0 Mobile & Touch (mobile/touch surfaces)

• UI-8.1 Hit target ≥44×44pt, 8px apart. The target may exceed the visible icon.
• UI-8.2 Primary actions in the bottom third — that is where the thumb lives.
• UI-8.3 Gestures are shortcuts, NEVER the only route. Anything swipeable is also tappable.
• UI-8.4 Respect safe areas and the keyboard. Notch, home indicator and on-screen keyboard all steal space.

9.0 Speed & Perceived Speed

• UI-9.1 Show structure before data — skeletons for known layouts, spinners only for unknown waits.
• UI-9.2 Assume success optimistically. Update UI first, reconcile after, roll back visibly on failure.
• UI-9.3 Reserve space for anything that loads. Images, ads and fonts must NEVER shift the layout.
• UI-9.4 Design for the slow connection. Offline, timeout and retry are states, NOT edge cases.

Apply by role — the clauses are one set; what you DO with them depends on the task:

Role	Obligation
DESIGN / PLAN a surface	Clauses shape the artifact: empty/loading/error specified first (UI-1.5), all 5 states enumerated (UI-5.2), type scale + spacing unit declared (UI-2.5, UI-4.1)
IMPLEMENT a component	Pre-completion gate: states · tokens · contrast · focus ring · touch target · reserved space (UI-5.2, UI-2.5/UI-4.1, UI-3.1, UI-5.5, UI-8.1, UI-9.3)
REVIEW UI code or a design artifact	Each clause is a fail-condition; every finding cites UI-<clause> + file:line + severity. NEVER a tick-box sweep — one focused pass per section
SCAN / document a UI system	Record where the PROJECT deliberately deviates, so the overriding doc becomes the recorded authority

Skip ONLY when the change has no user-facing surface (backend-only, tooling, docs) — state that explicitly so the skip is auditable, not an omission.

<!-- /SYNC:ui-ux-design-principles --> <!-- SYNC:understand-code-first:reminder -->

IMPORTANT MUST ATTENTION search 3+ existing patterns and read code BEFORE any modification. Run graph trace when graph.db exists.

<!-- /SYNC:understand-code-first:reminder --> <!-- SYNC:critical-thinking-mindset:reminder -->

MUST ATTENTION apply critical + sequential thinking — every claim needs appropriate traced evidence (file:line for repo/code claims; source URL or artifact section for research, product, content, and docs claims); confidence >80% to act, <60% DO NOT recommend. Anti-hallucination: never present guess as fact, admit uncertainty freely, cross-reference independently, stay skeptical of own confidence.

<!-- /SYNC:critical-thinking-mindset:reminder --> <!-- SYNC:ai-mistake-prevention:reminder -->

MUST ATTENTION apply AI mistake prevention — verify generated content against evidence, trace downstream references before deleting or renaming, verify all affected outputs, re-read files after context loss, and surface ambiguity before acting.

<!-- /SYNC:ai-mistake-prevention:reminder --> <!-- SYNC:task-tracking-external-report:reminder -->

• MANDATORY Bootstrap task tracking before target work; transition one task at a time.
• MANDATORY Persist plan/review findings to plans/reports/ incrementally and synthesize from disk.

<!-- /SYNC:task-tracking-external-report:reminder --> <!-- SYNC:project-reference-docs-guide:reminder -->

• MANDATORY Before investigating, planning, or coding, read docs/project-config.json (the project map: modules/paths, run-commands, conventions, architecture/workflow rules) + the required project-reference docs, and cite Reference docs read: ....
• MANDATORY Always include lessons.md; project config + conventions override generic framework defaults.
• MANDATORY If project config, root instruction files, or any required reference doc is missing or stale, auto-run /project-init or the narrow lower-level route before ordinary project-specific work.

<!-- /SYNC:project-reference-docs-guide:reminder --> <!-- SYNC:nested-task-creation:reminder -->

• MANDATORY Parent workflow rows do not replace child phase tracking; expand phases and link the parent when nested.
• MANDATORY Orchestrators pre-expand child skill phases before invocation; use [N.M] $skill-name — phase prefixes and one-in_progress discipline.

<!-- /SYNC:nested-task-creation:reminder --> <!-- SYNC:severity-rubric:reminder -->

• MANDATORY Classify findings Critical/High/Medium/Low by consequence; Critical/High block PASS until fixed or owner-accepted.
• MANDATORY Score-based skills (sre 0-2, perf two-axis) map onto the same four tiers — no parallel severity vocabulary.

<!-- /SYNC:severity-rubric:reminder --> <!-- SYNC:ui-intent-layer:reminder -->

• MANDATORY For UI-bearing specs, author/maintain the tech-agnostic interaction-surface layer (View Inventory + Navigation Map + observable UI States + per-story US-/OP-/BR--traced flow); keep deep visual fidelity in the linked design-spec/mockup recorded in frontmatter; name ZERO frameworks/routes/CSS/component classes; skip ONLY for backend-only features with a stated reason.

<!-- /SYNC:ui-intent-layer:reminder --> <!-- PROMPT-ENHANCE:STEP-TASK-CLOSING:START -->

Prompt-Enhance Closing Anchors

IMPORTANT MUST ATTENTION follow declared step order for this skill; NEVER skip, reorder, or merge steps without explicit user approval IMPORTANT MUST ATTENTION for every step/sub-skill call: set in_progress before execution, set completed after execution IMPORTANT MUST ATTENTION every skipped step MUST include explicit reason; every completed step MUST include concise evidence IMPORTANT MUST ATTENTION if Task tools unavailable, maintain an equivalent step-by-step plan tracker with synchronized statuses

<!-- PROMPT-ENHANCE:STEP-TASK-CLOSING:END --> <!-- SYNC:double-round-trip-review:reminder -->

• MANDATORY IMPORTANT MUST ATTENTION execute the review loop (aka Self-Review Convergence Loop): review → validate findings → fix validated findings → full re-review. A complete review pass with zero findings ENDS the review. Any newly produced output/judgment gets ≥1 self-review; any new judgment gets ≥1 /why-review --validate-findings pass before it is treated as final.
• MANDATORY apply the severity floor: rounds 1-2 exit on zero findings at any severity; from round 3 the bar is zero CRITICAL/HIGH/MEDIUM — LOW findings are no longer required to be fixed, so a LOW-only round ENDS the loop. List every deferred LOW in the report; NEVER re-tier a real CRITICAL/HIGH/MEDIUM down to LOW to reach the exit, and NEVER apply the floor to a binary gate (test-green, security must-fix).
• MANDATORY enforce the round cap of 3 — a ceiling, NEVER a target: a clean pass ends the loop immediately at any round (round 1 included), and round 3 completing with CRITICAL/HIGH/MEDIUM still open → STOP & escalate via AskUserQuestion, never a silent PASS. The 2-repeated-no-progress blocker rule is an earlier exit — escalate at whichever trips first. NEVER loop open-ended.

<!-- /SYNC:double-round-trip-review:reminder --> <!-- SYNC:goal-contract-satisfaction-loop:reminder -->

• MANDATORY Resolve the active Goal Contract BEFORE work (active plan goal.md → plans/goals/{YYMMDD-HHmm}-{slug}/goal.md → create from current request) and read saved success criteria before editing.
• MANDATORY Append iteration evidence after execution; emit a Goal Satisfaction matrix (PASS/FAIL/BLOCKED) before reporting PASS; loop on validated FAIL; escalate repeated no-progress or blockers. NEVER store secrets in goal files.

<!-- /SYNC:goal-contract-satisfaction-loop:reminder --> <!-- SYNC:trade-off-interrogation-gate:reminder -->

• MANDATORY MUST ATTENTION ALWAYS ASK THE 3 TRADE-OFF QUESTIONS — on the thing under review AND on every recommendation you make: (1) is there any trade-off? name what it SACRIFICES (change cost · complexity · perf · coupling · reversibility · migration · ops load · blast radius · security · testability · delivery time · UX) — "none"/"pure win" is an unfinished analysis, so state the dimensions checked; (2) is it worth it? gain (with a metric) vs cost, WHO pays, WHEN → emit WORTH IT / NOT WORTH IT / UNCLEAR; NOT WORTH IT → withdraw or replace it; (3) is it material enough to confirm with the user? irreversible/one-way door · cost shifted onto another team/ops/maintainer/user · one quality attribute traded for another · a tier/service/event/library boundary crossed · auth/money/data-integrity/breaking-change/High-or-Medium-risk path · verdict UNCLEAR → STOP and confirm via AskUserQuestion BEFORE the verdict.
• MANDATORY A MATERIAL trade-off with no user confirmation can NEVER be PASS; NEVER bury one as a Low-severity note, NEVER decide it silently, and NEVER let delivery or convergence pressure authorize a one-way door. — why: an un-walked-back one-way door is the user's call to make, not the reviewer's.
• MANDATORY — non-asking contexts escalate BY HANDOFF, never by silence. AskUserQuestion reaches only the main interactive agent: a sub-agent cannot ask the user, and a terminal/verdict-only mode asks nothing by design. There the duty is REDIRECTED, not waived — still name the trade-off, still decide materiality, record confirmed? = NO — cannot ask from this context, state the unconfirmed MATERIAL trade-off in your RETURNED verdict/summary so the CALLER escalates it (a note only in an on-disk report is not a handoff), and never emit an unqualified PASS. Applies ONLY where the user is genuinely unreachable (spawned sub-agent, terminal validate mode, headless run) — if you CAN ask, you MUST ask.

<!-- /SYNC:trade-off-interrogation-gate:reminder --> <!-- SYNC:parallel-subagent-dispatch -->

Parallel Sub-Agent Dispatch — Plan parallelism the moment a task breakdown exists, BEFORE executing it — running provably independent tasks sequentially wastes wall-clock. Applies to every multi-step job: workflow steps, planning, batch updates, investigation, research, scans, reviews, doc sync. Plan execution is metadata-gated, NEVER default-parallel — fan-out follows ONLY what the plan declares (PAR/SEQ tags + per-phase write set); an untagged plan runs sequentially — why: a derived write set cannot see cascade or generated writes.

1. Tag every task PAR or SEQ. PAR = inputs exclude every pending task's output AND write set disjoint from every other PAR. Else SEQ — MUST ATTENTION name the dependency forcing it.
2. Group PAR into waves. No edge between members. Two writers of one file NEVER share a wave. Read-only work (search, investigation, review, research) parallelizes freely.
3. Declare before dispatch: Parallel plan: wave 1 = [...] · wave 2 = [...] · SEQ = [...] (reason).
4. Spawn each wave in ONE message — every Agent call in one response, NEVER dripped per turn. Route each task to its specialist (.claude/skills/shared/sub-agent-selection-guide.md); NEVER code-reviewer as catch-all.
5. Brief each sub-agent self-contained: goal · scope + owned files · reference docs · return contract (summary + Full report: path, per SYNC:subagent-return-contract) · incremental persistence to plans/reports/ (per SYNC:incremental-persistence).
6. Barrier per wave. Advance ONLY after EVERY member returns (a skipped conditional counts as returned). Merge, mark each task completed/skipped, THEN dispatch the next wave. Mutating steps wait for the barrier.
7. One level deep. A dispatched sub-agent executes its own brief; further fan-out stays the orchestrator's job unless that agent's .claude/agents/*.md definition authorizes it.

NEVER parallelize: tasks sharing a write target · a task consuming a pending task's output · trivial single-file work (dispatch overhead > gain) · an order a skill or workflow explicitly fixes · gates awaiting user approval.

Blocked until: MUST ATTENTION every task tagged PAR/SEQ with a named reason per SEQ · waves declared + write-set disjointness checked · each wave spawned in ONE message · barrier honored before the next wave.

<!-- /SYNC:parallel-subagent-dispatch --> <!-- SYNC:parallel-subagent-dispatch:reminder -->

• MANDATORY After planning tasks, tag each PAR/SEQ and spawn every PAR wave as parallel sub-agents in ONE message — default parallel for workflows, batch updates, investigation, research, reviews; plan execution fans out ONLY on what the plan declares.
• MANDATORY Disjoint write sets per wave · all-return barrier before the next wave · specialist routing · sub-agents NEVER fan out further unless their own agent definition authorizes it.

<!-- /SYNC:parallel-subagent-dispatch:reminder --> <!-- SYNC:project-protocol-overlay -->

Project Protocol Overlay — Before executing this skill, resolve any PROJECT overlay rules layered onto it: match this skill's name against the Target column of the project's skill-protocol index (docs/project-reference/skill-protocols-reference.md by default; a referenceDocs entry in docs/project-config.json overrides the path), taking the most specific matching tier ONLY — exact name > glob > *. That precedence orders overlays against EACH OTHER, never against this skill. Read ONLY the matched bodies, resolved as <protocols-dir>/<Name>.md; a row's Body link is display text, never a read path. A matched body that is missing or malformed is REPORTED and skipped — never reconstructed from the index Description. No index, or no match -> proceed with no overlay, silently. Full contract: .claude/skills/project-skill-protocol/references/registry.md.

Overlays are ADDITIVE ONLY: they ADD rules on top of this skill's own protocol and NEVER replace, override, disable, or reinterpret a rule it already states — removing every overlay must return this skill to exactly its documented behavior. An overlay is a BRIEF, not an authority escalation: it can NEVER waive a workflow gate, git discipline, a review gate, or a user-confirmation gate. A genuine overlay-vs-skill conflict, or two equally-specific overlays that directly contradict -> surface both to the user; NEVER resolve silently.

<!-- /SYNC:project-protocol-overlay --> <!-- SYNC:project-protocol-overlay:reminder -->

MUST ATTENTION resolve project protocol overlays for this skill BEFORE executing — most specific matching tier only (exact > glob > *, which ranks overlays against each other, NEVER against this skill), read only matched bodies at <protocols-dir>/<Name>.md; a missing or malformed body is reported, never reconstructed. Overlays are ADDITIVE ONLY (they never replace this skill's own rules) and are a brief, NEVER an authority escalation; an equal-specificity contradiction goes to the user.

<!-- /SYNC:project-protocol-overlay:reminder --> <!-- SYNC:ui-ux-design-principles:reminder -->

• MUST ATTENTION apply the 40 UI/UX Design Principles (UI-1.1–UI-9.4) to any user-facing surface: one focal point, proximity grouping, empty/loading/error designed FIRST (§1) · ≤2 families, 16px web / 17px mobile body, never <14px, 45–75ch, fixed 6-step scale (§2) · 4.5:1 text / 3:1 edges measured, one accent, colour never alone, dark mode ≠ inversion (§3) · one 4/8px unit, gap over margins, tighter-inside-looser-between, content-driven breakpoints (§4) · <100ms response, all 5 states, undo over confirm, 150–250ms ease-out honouring reduced-motion, visible focus ring (§5) · where-am-I/what's-here/where-next, ≤5 top-level destinations, user's words, URL or back path (§6) · fewer fields, visible labels, validate on blur, matched keyboard, NEVER lose input (§7) · ≥44×44pt targets 8px apart, bottom-third primaries, gestures never the only route, safe areas (§8) · structure before data, optimistic update with visible rollback, reserved space, slow-connection states (§9). Project design-system docs OUTRANK these clauses — a genuine conflict goes to the user, NEVER resolved silently. Cite every finding as UI-<clause> + file:line. Skip ONLY for changes with no user-facing surface, stated explicitly.

<!-- /SYNC:ui-ux-design-principles:reminder -->

Closing Reminders

IMPORTANT MUST ATTENTION Goal: Review an artifact (PBI, design spec, story, test spec) for completeness and quality so reviewed artifacts are complete, evidence-backed, and ready for handoff without missing assumptions or acceptance gaps.

Protocols in force — MUST ATTENTION honor every block below (concise digest of the SYNC/shared blocks this skill carries):

• Nested Task Creation: Parent workflow rows never replace child phase tracking.
• Project Reference Docs Guide: Read required project docs before target work.
• Task Tracking External Report: Bootstrap tasks; persist review findings to plans/reports/.
• Critical Thinking Mindset: Traced file:line proof; confidence >80% to act.
• Evidence Based Reasoning: No claim without cited evidence; state confidence.
• Understand Code First: Read code, grep 3+ patterns before any change.
• Double Round Trip Review: Validate findings, fix, restart full review until clean.
• Fresh Context Review: Spawn fresh zero-memory sub-agent after each fix cycle.
• Review Protocol Injection: Embed all 11 protocol bodies verbatim in sub-agent prompts.
• AI Mistake Prevention: verify generated content against evidence, trace downstream references, verify all affected outputs, re-read after context loss, surface ambiguity.
• Severity Rubric: Classify findings Critical/High/Medium/Low by consequence.
• Parallel Sub-Agent Dispatch: Tag tasks PAR/SEQ, group PAR into disjoint-write-set waves, spawn each wave in ONE message, barrier before advancing.

IMPORTANT MUST ATTENTION execute the main steps IN ORDER — (1) Identify type → (2) Adversarial Mindset + Anti-Bias Gate → (3) type Required/Recommended checklist → (4) M1-M7 BLOCKING gate → (5) Readability checklist → (6) per-type output template → (7) validated-fix + full re-review loop until clean; NEVER skip or merge steps without explicit user approval — why: each step catches a distinct defect class the others miss. IMPORTANT MUST ATTENTION be a SKEPTIC, not a presence-checker — run ALL 6 adversarial techniques (steel-man rejected alternatives, stress-test 3 assumptions, AC-testability, pre-mortem, unseen alternatives, contrarian pass) and clear the Anti-Bias Gate BEFORE any verdict — why: sections that exist but hold weak/untestable content create false confidence worse than missing ones. IMPORTANT MUST ATTENTION enforce the BLOCKING M1-M7 gate on ALL types — any M1-M5 or M7 violation forces NEEDS WORK citing the mandate ID + exact section/line; NEVER pass an M1-M5/M7 violation — why: passing it makes this review itself defective. IMPORTANT MUST ATTENTION M7 (business-visibility) is judged on each case's BODY via the demo test — "what would a stakeholder SEE change?"; no answer → NEEDS WORK as TECHNICAL-ONLY. A When that is an invocation (handler runs, consumer receives, job fires, data syncs, model inspected) or a Then asserting schema/type/nullability/call-count FAILS M7 even in flawless tech-free prose — why: M1 governs vocabulary, M7 governs subject matter, and passing a case because its prose is clean is exactly how technical cases accumulate in business specs one bugfix at a time. IMPORTANT MUST ATTENTION exempt source identifiers inside evidence carriers ([Source:], **Evidence**, CoveredBy, legacy IntegrationTest, frontmatter, mermaid) — flag tech leakage only in narrative/AC/scenario prose — why: carriers are CORRECT places for class/path/test names; flagging them is a false finding. IMPORTANT MUST ATTENTION dispatch on --type={pbi|story|spec-tests|design} (infer if omitted) — apply that type's Required/Recommended checklist; verdict = PASS (all Required + ≥50% Recommended) | WARN (all Required, <50% Recommended) | FAIL (any Required fails). IMPORTANT MUST ATTENTION for --type=design ONLY: run the 9-dimension UI/UX Design Principles pass — all 40 clauses (UI-1.1-UI-9.4), one dimension at a time; every finding cites UI-<clause> + file:line + SYNC:severity-rubric severity and folds into the existing Required/Recommended verdict; pbi/story/spec-tests are unaffected, and project design-system docs OUTRANK the clauses. IMPORTANT MUST ATTENTION for --type=spec-tests: every [HARD] §4 rule / §5 invariant maps to ≥1 universally-quantified property TC + boundary counter-case — a missing property category is a blocking finding; one business TC covered by MANY tests is the correct one-to-many shape, NEVER a duplicate. IMPORTANT MUST ATTENTION run the findings-validation gate BEFORE fixing — invoke /why-review --validate-findings <report-path> first; NEVER edit the artifact to resolve findings before this gate returns CLEAN — why: validate-before-fix at parity with /plan-review prevents fixing phantom findings. IMPORTANT MUST ATTENTION fix only validated findings, then restart the FULL review with a fresh general-purpose sub-agent (artifacts are NOT code) and loop until a clean pass — a clean review with zero findings ENDS the loop; NEVER spawn a confirmation sub-agent after a clean round — why: every fix invalidates the prior verdict, but a clean pass needs no re-confirmation. IMPORTANT MUST ATTENTION cite file:line/section+line evidence for every finding (confidence >80% to act, <60% DO NOT recommend); every NEEDS WORK item must be actionable — why: speculation produces non-fixable findings. IMPORTANT MUST ATTENTION break work into small todo tasks using TaskCreate BEFORE starting; add a final review todo task to verify work quality.

[TASK-PLANNING] Before acting, analyze task scope and systematically break it into small todo tasks and sub-tasks using TaskCreate.

[IMPORTANT] Analyze how big the task is and break it into many small todo tasks systematically before starting — this is very important.

---

Closing reminder — Easy to Change is the success metric. Every finding, test, refactor, and abstraction must answer one question: does this make the next change cheaper or more expensive? If it doesn't reduce future change cost, reject it. Coupling, hidden state, duplicated knowledge, and unclear intent are the real enemies — call them out by name.

Anti-Rationalization:

Evasion	Rebuttal
"Required sections present, looks complete"	Presence ≠ quality. Name what's IN them and the specific failure mode NOT addressed.
"ACs are defined"	Are they TESTABLE? Name the automated test a QA engineer writes for each — without clarification.
"Alternatives were considered"	Real alternatives or strawmen set up to lose? Steel-man the strongest rejected one.
"Verdict is clear, skip the contrarian pass"	Generate 2 sentences arguing the OPPOSITE conclusion first, then decide on evidence.
"M1-M5/M7 violation is minor, let it pass"	Passing an M1-M5/M7 violation makes THIS review defective. NEEDS WORK + cite mandate ID + section/line.
"No tech words in it — M7 passes"	M1 ≠ M7. Apply the demo test to the BODY: what would a stakeholder SEE change? No answer → FAIL, however clean the prose.
"This sync/consumer case is business-critical, so it stays"	If it's business-critical it's demoable — rewrite it demoably. A case that CANNOT be rewritten demoably is exactly what M7 moves out.
"Source name in prose, flag it"	Check the carrier first — [Source:]/**Evidence**/CoveredBy/legacy IntegrationTest/frontmatter/mermaid are EXEMPT.
"Fix the finding, then I'm done"	Validate findings (/why-review) BEFORE fixing, then restart the FULL review until a clean pass.
"Skip evidence for review judgments"	Cite section+line for every finding; confidence >80% to act, <60% DO NOT recommend.

IMPORTANT MUST ATTENTION SKEPTIC stance — clear the Anti-Bias Gate (adversarial techniques) before any verdict. IMPORTANT MUST ATTENTION M6 enforcement — NEEDS WORK on any M1-M5 or M7 violation, cite mandate ID + section/line; carriers exempt. IMPORTANT MUST ATTENTION M7 — apply the demo test to each case's BODY, not its prose; an invocation-shaped When or a schema/type/call-count Then is TECHNICAL-ONLY and FAILS even when perfectly tech-free. IMPORTANT MUST ATTENTION validate findings before fixing, then restart the FULL fresh review until a clean pass ENDS the loop.