Przejdź do treści
Biblioteka skilli
Dane i analityka · Przegląd i audyt

review-database

/review-database

Przegląda bazy danych: SQL, migracje, bezpieczeństwo migracji, schemat, wydajność zapytań, indeksy, pule połączeń i transakcje w PostgreSQL i MySQL, także z użyciem squawk.

Repozytorium
skill-issue ↗
Licencja
MIT
Ostatnia zmiana repozytorium
16.10.2026

Ocena i bezpieczeństwo

100/ 100Klasa: dobry
Format25 / 25
Frontmatter z nazwą i opisem, poprawna nazwa, opis do 1024 znaków, nazwa zgodna z folderem.
Odkrywalność25 / 25
Opis mówi, kiedy użyć skilla, ma sensowną długość i jest w 3. osobie.
Treść35 / 35
Co najmniej 150 słów i do 500 linii, nagłówki, listy kroków, przykłady albo kod.
Higiena15 / 15
Bez zaślepek typu TODO, bez ścieżek Windows, z licencją.
Ryzykobrak sygnałów w skanie
Pliki w paczce
3
Skrypty
0
Komentarze HTML
1
Adresy URL w SKILL.md
1
Przestarzałość

Brak sygnałów. Repozytorium zmienione 16.10.2026, bez odwołań do wycofanych modeli i API.

Licencja

MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.

Co sprawdzić przed instalacją
  • SKILL.md zawiera komentarze HTML (1). Nie widać ich w podglądzie Markdown, a model je czyta.

Co robi i kiedy

Przegląda bazy danych: SQL, migracje, bezpieczeństwo migracji, schemat, wydajność zapytań, indeksy, pule połączeń i transakcje w PostgreSQL i MySQL, także z użyciem squawk.

Wyzwalacze z opisu

is this migration safewill this lock the table
Oryginalny opis (EN)

Use when the user asks for a database review, SQL review, migration review, migration safety check, schema review, schema design review, query performance review, index review, connection pool review, transaction management review, "is this migration safe", "will this lock the table", or a squawk run. Applies to PostgreSQL and MySQL. Also use when another skill dispatches a database review.

Jak działa

  1. Workflow
  2. Run metadata header
  3. Finding link wrapping (PR mode)
  4. Output Templates
  5. Guidelines
Sekcje z nagłówków SKILL.md, w oryginale · 128 linii, 1 072 słów
Podgląd SKILL.md
---
name: review-database
description: Use when the user asks for a database review, SQL review, migration review, migration safety check, schema review, schema design review, query performance review, index review, connection pool review, transaction management review, "is this migration safe", "will this lock the table", or a squawk run. Applies to PostgreSQL and MySQL. Also use when another skill dispatches a database review.
---

# Database & SQL Review

Structured database review producing actionable, prioritized findings with code-level references.

## Workflow

### 1. Scope and explore

- Confirm scope with the user: full codebase, specific packages/directories, specific migrations, changed files only (PR or branch diff), or specific concern.
- **Resolve scope to a file/package list.** Based on what the user requested:
  - **Changed files (PR or branch):** Run `git diff --name-only --diff-filter=d <base>...HEAD` to get changed files (default `<base>` is `main`). If the user references a PR number, use `gh pr diff <number> --name-only` instead. Filter to relevant file types (`.go`, `.sql`). Derive affected Go packages from the file paths (unique parent directories containing `.go` files).
  - **Explicit paths/packages:** The user may specify directories (e.g. `internal/store/`), Go package patterns (e.g. `./internal/store/...`), individual files, or specific migration files. When given a directory or package pattern, include all files under it.
  - **Full codebase:** No filtering. Explore everything (default).
- **Pass the resolved scope** (file list) to all exploration and investigation subagents so findings stay scoped to those files. Reading is **not** scoped: follow callees, callers, and third-party source wherever a finding depends on their behavior. Report bugs in code the scope touches or calls even when they predate the diff, prefixed `pre-existing:`.
- Explore the scoped code using parallel subagents (`subagent_type="explore"`). Read all in-scope `.sql` files, schema definitions, and source files that interact with the database (queries, connection setup, transaction handling).

### 2. Launch investigation subagent

Launch a single investigation subagent (`subagent_type="generalPurpose"`, `model: sonnet` per `subagent-model-routing`) with the list of in-scope files.

Prompt it to:
- Read all in-scope `.sql` files, schema definitions, and source files that interact with the database.
- Detect which database engine is in use by checking imports (`pgx`, `pq`, `lib/pq` for PostgreSQL; `go-sql-driver/mysql` for MySQL), migration tool configs, and connection strings. Apply the appropriate engine-specific checklist from [reference.md](reference.md).
- Analyze against all database categories: migration safety, query performance, connection & transaction management, and schema design (see [reference.md](reference.md)).
- Detect if horizontal sharding is in use (VSchema files, Citus distribution config, Spanner interleaved tables, shard-routing middleware, or multi-column partition keys). If detected, analyze hot-path queries against the checklist in [reference-sharding.md](reference-sharding.md): flag scatter queries, missing sharding keys in WHERE clauses, cross-shard JOINs, and cross-shard transactions.
- For PostgreSQL projects: check if `npx` is on PATH (`which npx`). If available, run `npx squawk-cli <migration_files>` against in-scope `.sql` migration files only (when scope is narrowed, limit to `.sql` files in the resolved file list). If `npx` is not available, check for `squawk` directly on PATH. If neither is available, note in Tool Availability and continue with manual review only.
- Include a **Tool Availability** section listing each tool's status (ran / skipped + reason).
- For each finding, search nearby code and project documentation for existing TODOs or notes.
- Return findings using the **per-category findings** template with `DB-` prefixed IDs (e.g. `DB1`, `DB2`).
- Every finding must include specific file paths, line numbers or function names, a severity rating (CRITICAL / HIGH / MEDIUM / LOW), and tracking status.

### 3. Present results

Resolve the review output directory (skip if `REVIEW_DIR` was provided by the review-all orchestrator):

```sh
REVIEW_DATE=$(date +%Y-%m-%d)
REVIEW_DIR=".reviews/${REVIEW_DATE}"
if [ -d "$REVIEW_DIR" ]; then REVIEW_DIR=".reviews/${REVIEW_DATE}-$(date +%H%M)"; fi
~/.claude/scripts/ensure-gitignore.sh '.reviews/'
mkdir -p "$REVIEW_DIR"
```

Capture run metadata (see [Run metadata header](#run-metadata-header) below) and prepend the rendered block to `${REVIEW_DIR}/DATABASE-REVIEW.md`.

Write the output to `${REVIEW_DIR}/DATABASE-REVIEW.md`, structured as:
1. Run metadata header
2. Tool availability summary
3. Findings table (with tracking status inline)
4. Recommended fix order

Present the report to the user.

---

## Run metadata header

Capture once near `REVIEW_DIR` resolution and prepend the rendered block to the output document:

```sh
RUN_DATETIME=$(date -u +"%Y-%m-%d %H:%M UTC")
GIT_BRANCH=$(git rev-parse --abbrev-ref HEAD)
GIT_COMMIT=$(git rev-parse --short HEAD)
GIT_COMMIT_FULL=$(git rev-parse HEAD)
GIT_SUBJECT=$(git log -1 --pretty=%s)
# When scope is diff-based, also: BASE_REF=<base>; BASE_COMMIT=$(git rev-parse --short "$BASE_REF")
```

Header template (placed at the top of the output `.md`, before the H1 title):

```markdown
> **Run:** {RUN_DATETIME}
> **Branch:** {GIT_BRANCH} @ {GIT_COMMIT} (`{GIT_COMMIT_FULL}`)
> **Subject:** {GIT_SUBJECT}
> **Base:** {BASE_REF} @ {BASE_COMMIT}   <!-- omit when scope is not diff-based -->
> **Scope:** {scope description}
```

---

## Finding link wrapping (PR mode)

When the review is scoped to a GitHub PR (`pr_url` is provided by the caller, or, when run standalone, `gh pr view --json url -q .url 2>/dev/null` returns one), wrap every `path:line` reference inside the finding tables below as a Markdown link:

```sh
~/.claude/scripts/pr-deeplink.sh "$pr_url" <path> <line>
# pr_url set   → [path:line](https://github.com/.../pull/N/files#diff-<hash>R<line>)
# pr_url empty → path:line   (plain text, unchanged)
```

The display text stays `path:line` so plain and linked tables look identical; only the URL goes in the link target. Pass `L` as the fourth argument for findings about removed code (default is `R`). Omit `<line>` for file-level findings to get a file-anchor link. Apply the same wrapping to `path:line` references inside the Tracked column (e.g. `TODO in foo.go:42`). Findings themselves follow `terse-comments`: concrete fix, optional `bug:`/`risk:`/`nit:`/`unsure:` prefix, no praise or restating the diff.

---

## Output Templates

### Per-category findings

```markdown
| # | Finding | Severity | Tracked |
|---|---------|----------|---------|
| DB1 | **Description.** Specific code reference (file:line). Explanation. | HIGH | — |
| DB2 | Description with code reference. | MEDIUM | TODO in file:line |
```

**Tracked column values:** Use `—` for new findings. For already-captured findings: `TODO in file:line`, `FIXME in file:line`, `README`, `#123` (issue reference), etc.

### Re-evaluation table (for follow-up reviews)

```markdown
| Finding | Status | What Changed |
|---------|--------|--------------|
| ~~1. Description~~ | FIXED | Brief explanation of the fix |
| 2. Description | Still applicable | No changes |
```

---

## Guidelines

- Search the organization's codebase (Sourcegraph, GitHub) for existing patterns before recommending changes.
- Include effort estimates to help prioritize implementation.
- When the user asks for a follow-up review, find the most recent review directory (`ls -d reviews/*/ 2>/dev/null | sort | tail -1`) containing `DATABASE-REVIEW.md`, re-evaluate all prior findings, and update with the re-evaluation table appended.
- For detailed framework categories, see [reference.md](reference.md).
- **REVIEW.md integration**: If a `REVIEW.md` context section was provided by the review-all orchestrator (or exists at the repository root when running standalone), treat its rules as additional review criteria. "Always check" items are HIGH severity; domain-specific items (Database section) are MEDIUM severity. "Skip" patterns exclude matching files from review scope.
- Findings must cite probed evidence (`path:line`, grep output, command result), not pattern-matched suspicion. Per `~/.claude/rules/probe-not-assume.md`.
Pliki w paczce (3)
  • SKILL.md 8 KB
  • reference-sharding.md 5 KB
  • reference.md 10 KB

Wersje

W bazie nie ma innych skilli o nazwie review-database.

Migawka bazy z 13.11.2026 · wersja paczki 8fb4f517c51f (skrót SHA-256 całej paczki)

Pobrania w ostatnich 30 dniach: 0 · ranking

Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.

Podobne skille

  • 100sql-expert
    Wspiera projektowanie, zapytania, optymalizację i administrację baz SQL w PostgreSQL, MySQL i SQL Server: indeksy, wydajność, transakcje i współbieżność.
  • 100postgresql
    Pomaga przy zmianach w PostgreSQL: schematy, SQL, indeksy, transakcje, migracje, role i wydajność bazy.
  • 95data-storage-migrations
    Projektuje schematy baz i migracje: wybór SQL lub NoSQL, indeksy, kopie zapasowe, retencję i cykl życia danych.
  • 86database-query
    Pisze zapytania SQL do zewnętrznych baz SQLite, MySQL i PostgreSQL, eksportuje wyniki i analizuje strukturę tabel.
  • 100fai-sql-code-review-skill
    Przegląda kod SQL pod kątem poprawności, wydajności, bezpieczeństwa i utrzymywalności w PostgreSQL, SQL Server i MySQL, w PR i skryptach migracji baz.
  • 90backend-database-patterns
    Pilnuje dobrych wzorców baz danych: projekt schematu, indeksy, problem N+1, granice transakcji, migracje i wzorzec repozytorium.

Uwagi i zgłoszenia

Wiadomość trafi do Bartosza Sokolińskiego. Adres e-mail jest potrzebny tylko do odpowiedzi. Szczegóły w polityce prywatności.