---

name: feature-arch-gate description: Adversarial pass/fail review gate that enforces the feature-arch skill's React feature-based architecture rules. Use when a diff, branch, PR, or src/ tree must be judged for architecture conformance before merge — feature folder structure, import boundaries, cross-feature isolation, data-fetching, state, testing, and naming rules. A single blind reviewer renders per-rule PASS/FAIL verdicts with cited evidence, and the reviewer's structured output is the verdict fail-closed. Use the feature-arch skill itself to design or migrate an architecture; use this gate to judge whether work conforms to it.

Feature-Arch Gate

Enforce React feature-based architecture conformance — a pass/fail gate: a single blind reviewer subagent judges the work against the rules of feature-arch v1.1.0 with an adversarial mandate, and the work passes only when every rule is PASS or N/A. This skill renders verdicts; it never fixes the work.

The gate is self-sufficient: the 33 decidable rules it enforces (of the source's 43) are vendored into this skill's references/ directory as a snapshot of feature-arch v1.1.0, so a review needs nothing outside this folder. references/_rule-evidence.md records the provenance, the evidence that decides each rule, and the 10 source rules excluded as non-decidable.

When to Apply

• A PR or branch touching a feature-organized React codebase needs an architecture conformance verdict before merge.
• A migration step from a FEATURE-ARCH-TARGET.md blueprint is claimed complete and needs independent verification.
• A full src/ tree audit is requested ("does this codebase conform to feature-arch?").
• Agent-generated feature work needs a gate that is blind to the conversation that produced it.

Do NOT apply this gate to design or migrate an architecture — that is the feature-arch skill's job (it produces the blueprint; this skill judges conformance to the rules).

Review Protocol

Follow these steps exactly — the gate's value is that every review runs the same way.

1. Identify the target. Pin down exactly what is under review (a diff, a set of files, or a full src/ tree) and note the ref/paths so the review runs against an unambiguous, fixed target. Record two context facts the reviewer needs: does the codebase use React Server Components, and is a query library (TanStack Query etc.) present — two rules are N/A without them.
2. Load the rules. Read references/_rule-evidence.md and resolve the 33 vendored rule files it lists against this skill's own references/ directory. If any listed rule file is missing or unreadable, STOP and report the error — never proceed with partial rules or silently pass.
3. Compose the reviewer prompt. Fill references/reviewer-prompt.md with the rule file paths, the target, and (if the target repo has one) its docs/architecture/FEATURE-ARCH-TARGET.md blueprint. The composed prompt must be fully self-contained — a reviewer sees no conversation history, so nothing may refer to context outside the prompt.
4. Dispatch one blind reviewer. Launch a single Task subagent whose entire input is the composed prompt — no conversation context, no commentary alongside it.
5. Render fail-closed. The reviewer's structured output is the verdict — there is no merge step. Overall verdict is PASS only when every rule is PASS or N/A; any single FAIL fails the gate. Never average, weigh severity, or waive a rule — a "minor" FAIL is a FAIL.
6. Render the verdict. Fill assets/templates/verdict.md. On FAIL, aggregate the reviewer's "missing for PASS" suggestions into the fix list, each with its location, ordered by the source skill's category priority (struct → import → bound → fquery → fcomp → fstate → test → name). Every rule whose final result is FAIL must appear in the fix list with a change concrete enough to apply as written — if the reviewer's suggestion only restates the violation, derive the fix from the rule's Correct example before rendering.

If the same rule flips verdicts across re-reviews of an unchanged target, or a human reads the evidence and overrides the verdict, that is a decidability bug in the rule — record it in gotchas.md and sharpen the rule; do not override the gate.

Verdict Format

The reviewer returns, per rule: PASS | FAIL | N/A, evidence (file:line or a quote — required for PASS as well as FAIL), and for every FAIL, the fix that flips the rule to PASS once applied — the named change plus its location, never a restatement of the violation. The final report follows assets/templates/verdict.md.

Rule Categories

The eight categories and their priority order are defined in references/_sections.md; the 33 vendored rule files live alongside it in references/, and references/_rule-evidence.md lists them with the evidence that decides each, plus the 10 excluded source rules (with reasons).

Reference Files

File	Description
references/reviewer-prompt.md	Self-contained prompt template for the blind reviewer
assets/templates/verdict.md	Verdict report template
references/_rule-evidence.md	Vendoring provenance + per-rule deciding evidence + exclusions
references/_sections.md	Category definitions and fix-list priority order
references/<category>-*.md	The 33 vendored rule files (struct/import/bound/fquery/fcomp/fstate/test/name)
metadata.json	Version and source references

Related Skills

• feature-arch — the source distillation skill this gate's rules were vendored from; use it (if available) to design or migrate the target architecture, and to read the 10 judgment-call rules the gate excludes. The gate itself never needs it at review time.