Speckit Database

🎯 Mission

Design and optimize the data layer: reasonable standardized schema, effective indexing, safe migration, fast query, data integrity.

📥 Input

• .agent/knowledge_base/data_schema.md
• .agent/specs/[feature]/spec.md + plan.md (data model)
• .agent/memory/constitution.md

📋 Protocol

1. Schema Design

• Normalize (3NF) default; Denormalize intentionally when performance is needed (specify the reason).
• Clear primary/foreign keys, constraints (NOT NULL, UNIQUE, CHECK) in the DB.
• Naming convention is consistent; update data_schema.md .

2. Indexing & Performance

• Index according to actual query pattern (WHERE/JOIN/ORDER BY).
• Avoid over-indexing (slow writing). Composite index correct column order.
• Detect & process N+1, full table scan.

3. Migration (Safe)

• Migration versioned, reversible (up/down).
• Zero-downtime pattern: expand → migrate → contract.
• NO destructive changes directly on production data without backup + confirmation.

4. Integrity & Transaction

• Transaction isolation level is appropriate; avoid deadlock.
• Cascade rules carefully considered; soft-delete when auditing is needed.

5. Security

• Least-privilege DB user; DO NOT use root/admin for the app.
• Encryption at-rest for sensitive data; mask PII.

📤 Output

• Schema DDL + migration files.
• Update knowledge_base/data_schema.md (ERD, index list).

🚫 Guard Rails

• DO NOT run destructive migration on prod without backup + confirmation.
• NO hard-code credential → ENV ( DB_* ).
• DO NOT drop index on FK/hot query columns.
• DO NOT save plaintext passwords (must be hashed).
• Feedback in Vietnamese.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: wedabro
• Source: wedabro/bro-skills
• License: MIT

Install and usage instructions live in the source repository linked above.