sql-database-assistant
Use when the user asks to write SQL queries, optimize database performance, generate migrations, explore database schemas, or work with ORMs like Prisma, Drizzle, TypeORM, or SQLAlchemy.
- License 许可
- MIT
- 最近更新
- 2026-09-22
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- sql-database-assistant
- description
- Use when the user asks to write SQL queries, optimize database performance, generate migrations, explore database schemas, or work with ORMs like Prisma, Drizzle, TypeORM, or SQLAlchemy.
02 资源文件 6
- 参考文档references/optimization_guide.md10.4 KB
- 参考文档references/orm_patterns.md12.3 KB
- 参考文档references/query_patterns.md9.6 KB
- 脚本scripts/migration_generator.py16.1 KB
- 脚本scripts/query_optimizer.py11.5 KB
- 脚本scripts/schema_explorer.py14.0 KB
03 SKILL.md 正文
SQL Database Assistant - POWERFUL Tier Skill
Overview
The operational companion to database design. While database-designer focuses on schema architecture and database-schema-designer handles ERD modeling, this skill covers the day-to-day: writing queries, optimizing performance, generating migrations, and bridging the gap between application code and database engines.
Core Capabilities
- Natural Language to SQL — translate requirements into correct, performant queries
- Schema Exploration — introspect live databases across PostgreSQL, MySQL, SQLite, SQL Server
- Query Optimization — EXPLAIN analysis, index recommendations, N+1 detection, rewrite patterns
- Migration Generation — up/down scripts, zero-downtime strategies, rollback plans
- ORM Integration — Prisma, Drizzle, TypeORM, SQLAlchemy patterns and escape hatches
- Multi-Database Support — dialect-aware SQL with compatibility guidance
Tools
| Script | Purpose |
|---|---|
scripts/query_optimizer.py |
Static analysis of SQL queries for performance issues |
scripts/migration_generator.py |
Generate migration file templates from change descriptions |
scripts/schema_explorer.py |
Generate schema documentation from introspection queries |
Natural Language to SQL
Translation Patterns
When converting requirements to SQL, follow this sequence:
- Identify entities — map nouns to tables
- Identify relationships — map verbs to JOINs or subqueries
- Identify filters — map adjectives/conditions to WHERE clauses
- Identify aggregations — map "total", "average", "count" to GROUP BY
- Identify ordering — map "top", "latest", "highest" to ORDER BY + LIMIT
Common Query Templates
Top-N per group (window function)
SELECT * FROM (
SELECT *, ROW_NUMBER() OVER (PARTITION BY department_id ORDER BY salary DESC) AS rn
FROM employees
) ranked WHERE rn <= 3;
Running totals
SELECT date, amount,
SUM(amount) OVER (ORDER BY date ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS running_total
FROM transactions;
Gap detection
SELECT curr.id, curr.seq_num, prev.seq_num AS prev_seq
FROM records curr
LEFT JOIN records prev ON prev.seq_num = curr.seq_num - 1
WHERE prev.id IS NULL AND curr.seq_num > 1;
UPSERT (PostgreSQL)
INSERT INTO settings (key, value, updated_at)
VALUES ('theme', 'dark', NOW())
ON CONFLICT (key) DO UPDATE SET value = EXCLUDED.value, updated_at = EXCLUDED.updated_at;
UPSERT (MySQL)
INSERT INTO settings (key_name, value, updated_at)
VALUES ('theme', 'dark', NOW())
ON DUPLICATE KEY UPDATE value = VALUES(value), updated_at = VALUES(updated_at);
See references/query_patterns.md for JOINs, CTEs, window functions, JSON operations, and more.
Schema Exploration
Introspection Queries
PostgreSQL — list tables and columns
SELECT table_name, column_name, data_type, is_nullable, column_default
FROM information_schema.columns
WHERE table_schema = 'public'
ORDER BY table_name, ordinal_position;
PostgreSQL — foreign keys
SELECT tc.table_name, kcu.column_name,
ccu.table_name AS foreign_table, ccu.column_name AS foreign_column
FROM information_schema.table_constraints tc
JOIN information_schema.key_column_usage kcu ON tc.constraint_name = kcu.constraint_name
JOIN information_schema.constraint_column_usage ccu ON tc.constraint_name = ccu.constraint_name
WHERE tc.constraint_type = 'FOREIGN KEY';
MySQL — table sizes
SELECT table_name, table_rows,
ROUND(data_length / 1024 / 1024, 2) AS data_mb,
ROUND(index_length / 1024 / 1024, 2) AS index_mb
FROM information_schema.tables
WHERE table_schema = DATABASE()
ORDER BY data_length DESC;
SQLite — schema dump
SELECT name, sql FROM sqlite_master WHERE type = 'table' ORDER BY name;
SQL Server — columns with types
SELECT t.name AS table_name, c.name AS column_name,
ty.name AS data_type, c.max_length, c.is_nullable
FROM sys.columns c
JOIN sys.tables t ON c.object_id = t.object_id
JOIN sys.types ty ON c.user_type_id = ty.user_type_id
ORDER BY t.name, c.column_id;
Generating Documentation from Schema
Use scripts/schema_explorer.py to produce markdown or JSON documentation:
python scripts/schema_explorer.py --dialect postgres --tables all --format md
python scripts/schema_explorer.py --dialect mysql --tables users,orders --format json --json
Query Optimization
EXPLAIN Analysis Workflow
- Run EXPLAIN ANALYZE (PostgreSQL) or EXPLAIN FORMAT=JSON (MySQL)
- Identify the costliest node — Seq Scan on large tables, Nested Loop with high row estimates
- Check for missing indexes — sequential scans on filtered columns
- Look for estimation errors — planned vs actual rows divergence signals stale statistics
- Evaluate JOIN order — ensure the smallest result set drives the join
Index Recommendation Checklist
- Columns in WHERE clauses with high selectivity
- Columns in JOIN conditions (foreign keys)
- Columns in ORDER BY when combined with LIMIT
- Composite indexes matching multi-column WHERE predicates (most selective column first)
- Partial indexes for queries with constant filters (e.g.,
WHERE status = 'active') - Covering indexes to avoid table lookups for read-heavy queries
Query Rewriting Patterns
| Anti-Pattern | Rewrite |
|---|---|
SELECT * FROM orders |
SELECT id, status, total FROM orders (explicit columns) |
WHERE YEAR(created_at) = 2025 |
WHERE created_at >= '2025-01-01' AND created_at < '2026-01-01' (sargable) |
| Correlated subquery in SELECT | LEFT JOIN with aggregation |
NOT IN (SELECT ...) with NULLs |
NOT EXISTS (SELECT 1 ...) |
UNION (dedup) when not needed |
UNION ALL |
LIKE '%search%' |
Full-text search index (GIN/FULLTEXT) |
ORDER BY RAND() |
Application-side random sampling or TABLESAMPLE |
N+1 Detection
Symptoms:
- Application loop that executes one query per parent row
- ORM lazy-loading related entities inside a loop
- Query log shows hundreds of identical SELECT patterns with different IDs
Fixes:
- Use eager loading (
includein Prisma,joinedloadin SQLAlchemy) - Batch queries with
WHERE id IN (...) - Use DataLoader pattern for GraphQL resolvers
Static Analysis Tool
python scripts/query_optimizer.py --query "SELECT * FROM orders WHERE status = 'pending'" --dialect postgres
python scripts/query_optimizer.py --query queries.sql --dialect mysql --json
See references/optimization_guide.md for EXPLAIN plan reading, index types, and connection pooling.
Migration Generation
Zero-Downtime Migration Patterns
Adding a column (safe)
-- Up
ALTER TABLE users ADD COLUMN phone VARCHAR(20);
-- Down
ALTER TABLE users DROP COLUMN phone;
Renaming a column (expand-contract)
-- Step 1: Add new column
ALTER TABLE users ADD COLUMN full_name VARCHAR(255);
-- Step 2: Backfill
UPDATE users SET full_name = name;
-- Step 3: Deploy app reading both columns
-- Step 4: Deploy app writing only new column
-- Step 5: Drop old column
ALTER TABLE users DROP COLUMN name;
Adding a NOT NULL column (safe sequence)
-- Step 1: Add nullable
ALTER TABLE orders ADD COLUMN region VARCHAR(50);
-- Step 2: Backfill with default
UPDATE orders SET region = 'unknown' WHERE region IS NULL;
-- Step 3: Add constraint
ALTER TABLE orders ALTER COLUMN region SET NOT NULL;
ALTER TABLE orders ALTER COLUMN region SET DEFAULT 'unknown';
Index creation (non-blocking, PostgreSQL)
CREATE INDEX CONCURRENTLY idx_orders_status ON orders (status);
Data Backfill Strategies
- Batch updates — process in chunks of 1000-10000 rows to avoid lock contention
- Background jobs — run backfills asynchronously with progress tracking
- Dual-write — write to old and new columns during transition period
- Validation queries — verify row counts and data integrity after each batch
Rollback Strategies
Every migration must have a reversible down script. For irreversible changes:
- Backup before execution —
pg_dumpthe affected tables - Feature flags — application can switch between old/new schema reads
- Shadow tables — keep a copy of the original table during migration window
Migration Generator Tool
python scripts/migration_generator.py --change "add email_verified boolean to users" --dialect postgres --format sql
python scripts/migration_generator.py --change "rename column name to full_name in customers" --dialect mysql --format alembic --json
Multi-Database Support
Dialect Differences
| Feature | PostgreSQL | MySQL | SQLite | SQL Server |
|---|---|---|---|---|
| UPSERT | ON CONFLICT DO UPDATE |
ON DUPLICATE KEY UPDATE |
ON CONFLICT DO UPDATE |
MERGE |
| Boolean | Native BOOLEAN |
TINYINT(1) |
INTEGER |
BIT |
| Auto-increment | SERIAL / GENERATED |
AUTO_INCREMENT |
INTEGER PRIMARY KEY |
IDENTITY |
| JSON | JSONB (indexed) |
JSON |
Text (ext) | NVARCHAR(MAX) |
| Array | Native ARRAY |
Not supported | Not supported | Not supported |
| CTE (recursive) | Full support | 8.0+ | 3.8.3+ | Full support |
| Window functions | Full support | 8.0+ | 3.25.0+ | Full support |
| Full-text search | tsvector + GIN |
FULLTEXT index |
FTS5 extension | Full-text catalog |
| LIMIT/OFFSET | LIMIT n OFFSET m |
LIMIT n OFFSET m |
LIMIT n OFFSET m |
OFFSET m ROWS FETCH NEXT n ROWS ONLY |
Compatibility Tips
- Always use parameterized queries — prevents SQL injection across all dialects
- Avoid dialect-specific functions in shared code — wrap in adapter layer
- Test migrations on target engine —
information_schemavaries between engines - Use ISO date format —
'YYYY-MM-DD'works everywhere - Quote identifiers — use double quotes (SQL standard) or backticks (MySQL)
ORM Patterns
Prisma
Schema definition
model User {
id Int @id @default(autoincrement())
email String @unique
name String?
posts Post[]
createdAt DateTime @default(now())
}
model Post {
id Int @id @default(autoincrement())
title String
author User @relation(fields: [authorId], references: [id])
authorId Int
}
Migrations: npx prisma migrate dev --name add_user_email
Query API: prisma.user.findMany({ where: { email: { contains: '@' } }, include: { posts: true } })
Raw SQL escape hatch: prisma.$queryRaw\SELECT * FROM users WHERE id = ${userId}``
Drizzle
Schema-first definition
export const users = pgTable('users', {
id: serial('id').primaryKey(),
email: varchar('email', { length: 255 }).notNull().unique(),
name: text('name'),
createdAt: timestamp('created_at').defaultNow(),
});
Query builder: db.select().from(users).where(eq(users.email, email))
Migrations: npx drizzle-kit generate:pg then npx drizzle-kit push:pg
TypeORM
Entity decorators
@Entity()
export class User {
@PrimaryGeneratedColumn()
id: number;
@Column({ unique: true })
email: string;
@OneToMany(() => Post, post => post.author)
posts: Post[];
}
Repository pattern: userRepo.find({ where: { email }, relations: ['posts'] })
Migrations: npx typeorm migration:generate -n AddUserEmail
SQLAlchemy
Declarative models
class User(Base):
__tablename__ = 'users'
id = Column(Integer, primary_key=True)
email = Column(String(255), unique=True, nullable=False)
name = Column(String(255))
posts = relationship('Post', back_populates='author')
Session management: Always use with Session() as session: context manager
Alembic migrations: alembic revision --autogenerate -m "add user email"
See references/orm_patterns.md for side-by-side comparisons and migration workflows per ORM.
Data Integrity
Constraint Strategy
- Primary keys — every table must have one; prefer surrogate keys (serial/UUID)
- Foreign keys — enforce referential integrity; define ON DELETE behavior explicitly
- UNIQUE constraints — for business-level uniqueness (email, slug, API key)
- CHECK constraints — validate ranges, enums, and business rules at the DB level
- NOT NULL — default to NOT NULL; make nullable only when genuinely optional
Transaction Isolation Levels
| Level | Dirty Read | Non-Repeatable Read | Phantom Read | Use Case |
|---|---|---|---|---|
| READ UNCOMMITTED | Yes | Yes | Yes | Never recommended |
| READ COMMITTED | No | Yes | Yes | Default for PostgreSQL, general OLTP |
| REPEATABLE READ | No | No | Yes (InnoDB: No) | Financial calculations |
| SERIALIZABLE | No | No | No | Critical consistency (billing, inventory) |
Deadlock Prevention
- Consistent lock ordering — always acquire locks in the same table/row order
- Short transactions — minimize time between first lock and commit
- Advisory locks — use
pg_advisory_lock()for application-level coordination - Retry logic — catch deadlock errors and retry with exponential backoff
Backup & Restore
PostgreSQL
# Full backup
pg_dump -Fc --no-owner dbname > backup.dump
# Restore
pg_restore -d dbname --clean --no-owner backup.dump
# Point-in-time recovery: configure WAL archiving + restore_command
MySQL
# Full backup
mysqldump --single-transaction --routines --triggers dbname > backup.sql
# Restore
mysql dbname < backup.sql
# Binary log for PITR: mysqlbinlog --start-datetime="2025-01-01 00:00:00" binlog.000001
SQLite
# Backup (safe with concurrent reads)
sqlite3 dbname ".backup backup.db"
Backup Best Practices
- Automate — cron or systemd timer, never manual-only
- Test restores — untested backups are not backups
- Offsite copies — S3, GCS, or separate region
- Retention policy — daily for 7 days, weekly for 4 weeks, monthly for 12 months
- Monitor backup size and duration — sudden changes signal issues
Anti-Patterns
| Anti-Pattern | Problem | Fix |
|---|---|---|
SELECT * |
Transfers unnecessary data, breaks on schema changes | Explicit column list |
| Missing indexes on FK columns | Slow JOINs and cascading deletes | Add indexes on all foreign keys |
| N+1 queries | 1 + N round trips to database | Eager loading or batch queries |
| Implicit type coercion | WHERE id = '123' prevents index use |
Match types in predicates |
| No connection pooling | Exhausts connections under load | PgBouncer, ProxySQL, or ORM pool |
| Unbounded queries | No LIMIT risks returning millions of rows | Always paginate |
| Storing money as FLOAT | Rounding errors | Use DECIMAL(19,4) or integer cents |
| God tables | One table with 50+ columns | Normalize or use vertical partitioning |
| Soft deletes everywhere | Complicates every query with WHERE deleted_at IS NULL |
Archive tables or event sourcing |
| Raw string concatenation | SQL injection | Parameterized queries always |
Cross-References
| Skill | Relationship |
|---|---|
| database-designer | Schema architecture, normalization analysis, ERD generation |
| database-schema-designer | Visual ERD modeling, relationship mapping |
| migration-architect | Complex multi-step migration orchestration |
| api-design-reviewer | Ensuring API endpoints align with query patterns |
| observability-platform | Query performance monitoring, slow query alerts |
04 同仓库更多 Skills
syllabus
Generates a curated supplementary reading list from any course syllabus using Consensus academic search. Grill-me intake (syllabus input format + course audience + year range) plus a grouping forcing-options checkpoint before any search runs — so the reading list matches the course's level and recency need. Parses the syllabus to extract topics and learning outcomes, searches Consensus for recent peer-reviewed papers per topic, and produces a professionally formatted .docx with clickable Consensus links, plain-language summaries calibrated to audience level, and Bloom-higher-order discussion questions tied to course learning goals. Triggers whenever a user uploads a syllabus, course outline, or curriculum document and wants supplementary readings. Also triggers on: 'syllabus reading list', 'find papers for my course', 'create a reading list from this syllabus', 'recent research for my class', 'supplementary readings', 'find journal articles for these topics', 'what recent papers cover this material', 'any new research on these course topics', 'update my syllabus with recent papers'. Even casual mentions when a syllabus is attached should trigger this skill.
research
Default entry point for any research request — a hybrid router that classifies the question deterministically and either delegates to a specialist research skill (pulse for trends/sentiment, grants for NIH funding, litreview for academic literature, syllabus for course reading, patent for prior-art + IP landscape, dossier for entity research) or runs its own plan-decompose-multi-source-search-synthesize-cite fallback workflow when no specialist matches. Always surfaces the routing decision so users can override. Triggers — "research [topic]", "look into [topic]", "what do we know about [topic]", "investigate [topic]", "find me information on [topic]", "do some research on [topic]", "I need to understand [topic]", or any research request that doesn't obviously match a more-specific specialist skill. Output is a markdown briefing (default) or .docx document (on request) with full citations and an audit log.
pulse
Multi-source recency research skill that takes the pulse of any topic across Reddit, Hacker News, the open web, and optionally X/Twitter within a configurable recent window (default 30 days). Forcing intake clarifies topic specificity, angle (trend/sentiment/problems/opportunities/comparison), time window, and platform scope before searching. Returns a synthesized briefing with citations, engagement metrics, and cross-platform pattern analysis. Triggers: 'pulse on [topic]', 'what's happening with [topic]', 'what are people saying about [topic]', 'current conversation about [topic]', 'take the pulse of [topic]', 'trending: [topic]', 'find me info on [topic]', or any variation requesting multi-source recency intelligence on a topic. Also use for competitor research, trend discovery, tool comparisons, and audience sentiment analysis.
patent
Patent prior-art and landscape intelligence skill — not generic patent help. Commits to one of five sub-use-cases via forcing intake (novelty search / freedom-to-operate / competitive landscape / acquisition diligence / litigation prior-art) before any search runs. Searches Google Patents, Espacenet, USPTO, and optionally Lens.org for citation-graph signals. Output is an editable Word document (.docx) with verdict, ranked closest art (claim-text extracted), CPC-class-aware landscape, family-resolved hits, geographic coverage, FTO flags where applicable, strategy recommendations, and full audit log. Triggers: 'prior art search for [invention]', 'patent search on [topic]', 'freedom to operate analysis', 'FTO for [product]', 'patent landscape for [field]', 'is [invention] novel', 'patents on [topic]', 'competitive patent analysis', 'prior art for litigation', 'patent diligence on [company]'. Produces search signal, not legal advice — always recommends consulting a patent attorney before filing or licensing decisions. Trademark, copyright, and trade-secret questions are out of scope.
notebooklm
Browser automation skill for controlling Google's NotebookLM. Handles reading and querying notebooks, adding sources (URLs, text, files, YouTube links, synthesized content), generating Studio outputs (Audio Overview, infographics, slide decks, study guides, briefing docs, mind maps, timelines, FAQs), and creating new notebooks. Triggers on any phrase involving NotebookLM — 'open NotebookLM', 'check my [name] notebook', 'pull info from NotebookLM', 'ask my notebook about X', 'add [source] to NotebookLM', 'create an infographic in NotebookLM', 'use NotebookLM Studio', 'generate a slide deck from my notebook', or any variation where the goal involves NotebookLM. Requires browser automation environment — fails gracefully when unavailable.
litreview
Academic literature orientation skill that searches papers via Consensus, builds a strategic search plan using PICO (default) or SPIDER / Decomposition / hybrid as fallbacks, and synthesizes findings into a professionally formatted Word document (.docx) research guide. Grill-me intake (research question specificity + framework hint + tentative depth) before the recon search; a second forcing checkpoint after Phase 2 confirms framework + sub-areas + depth before searches consume budget. Configurable depth (5/10/20 queries) controls coverage vs. speed. Output is a 'launching pad' — not a finished review, but an orientation guide that lets a researcher dive in confidently. Triggers: 'litreview on [topic]', 'literature review on [topic]', 'I'm starting a literature review on X', 'I'm writing a paper on X', 'help me research X', 'I'm doing research on X', 'can you help me research X'. Do NOT trigger for single one-off paper searches where the user just wants a quick list — that's a plain Consensus search.