Claude Skills by AppVerk

Use when resolving code-review findings flagged needs-decision in bulk — the analysis fan-out, the decision sweep and its five outcomes, the dispatch contract, orchestrator-run verification, and the decision record written into the report. Loaded by /fix-report and /fix-all; /fix loads it for the Alternatives render format alone.

Use when checking or reporting docs↔code drift — a declarative registry (claim → source of truth → policy) with a three-way classification: mechanical facts auto-fixable, judgment calls escalated, dead references flagged as inconsistencies whose fix is a decision.

Use when authoring or reviewing any code-review reporting agent, or when a code-review agent is about to report findings — mandates a self-falsification pass: every finding survives a refutation battery before reporting; rejected findings and doctrine-gap candidates are recorded in dedicated sections, never silently dropped.

Use when writing or reviewing the definition of a reporting agent or command (a review, audit, verification, or test-run agent), before its closing contract is finalized.

Bun package management, lockfile policy, workspaces, CI integration, and Bun-native tooling

Use when a component or view is driven by two or more independent boolean inputs (flags, permissions, connection states) — enumerate the full 2^N product, confirm which combinations are real, never collapse independent axes into one exclusive switch.

Use when designing, authoring, or reviewing a closed agent loop (test→fix→retest, audit→fix→re-audit, generate→verify→correct) in this marketplace — the minimum-bar checklist, the ground-truth oracle taxonomy, and the anti-patterns, anchored to /qa:loop as the reference implementation.

Use when authoring or reviewing a fan-out reader/scout agent that ingests an external source (a design board, a live site, a large document) on behalf of a consumer.

Lens roster, panel-selection rules, and severity/needs-decision anchors for the /superutils:spec-review loop. Load when composing a review panel or grading findings.

Report structure, sidecar schema, SR-id rules, outcome enum, and terminal statuses for the /superutils:spec-review loop. Load when reading or writing spec-review loop state or reports. (Distinct from qa:report-format, which is the QA test-report format.)

Backend testing patterns — API request construction, response verification, database state checks, error handling testing, and adaptive tool detection.

Frontend testing patterns using Playwright MCP — navigation, interaction, assertions, screenshots on failure, and common UI testing scenarios.

Test report format with QA-XXX issue IDs compatible with code-review plugin. Defines report structure, severity levels, issue format, and detailed results.