Database Skill

You now have read-only access to multiple databases through the database command. This skill supports connecting to production, staging, local, or any other configured database.

Important Safety Rules

1. READ-ONLY ACCESS: This skill only allows SELECT, SHOW, DESCRIBE, and EXPLAIN queries
2. NO MODIFICATIONS: You cannot INSERT, UPDATE, DELETE, or modify any data
3. QUERY VALIDATION: All queries are validated before execution
4. PRODUCTION DATA: Treat all data as sensitive and confidential
5. MULTIPLE DATABASES: You can configure and switch between multiple database connections

Available Commands

Managing Multiple Databases

database dbs

List all configured database connections with their connection status. Shows which database is currently active.

Using a specific database:

database --db production <command>
database --db staging query "SELECT * FROM tasks"
database --db local tables

If you don't specify --db, the default database will be used (usually the first one configured).

Connection Testing

database test

Test database connectivity and show connection details for the active database.

database --db staging test

Test connection to a specific database.

Database Exploration

database tables

List all tables in the database.

database describe <table_name>

Show the schema (columns, types) and indexes for a specific table.

database stats

Show database statistics including row counts for all tables.

Custom Queries

database query "SELECT * FROM tasks WHERE status = 'open' LIMIT 10"

Execute a custom SELECT query. Returns results in table format.

database query "SELECT COUNT(*) FROM users" json

Execute a query and return results in JSON format (useful for parsing).

Supported formats:

• table (default) - Human-readable table format
• json - JSON array of objects
• csv - Comma-separated values

Usage Guidelines

When to Use This Skill

Use database when you need to:

• Investigate data issues or bugs
• Answer questions about production data
• Generate reports or statistics
• Understand data relationships
• Debug API behavior by checking database state
• Validate data integrity

Best Practices

1. Start with exploration: Use tables and describe to understand the schema
2. Use LIMIT: Always limit results to avoid overwhelming output
3. Use predefined queries: When available, use shortcuts like database tasks instead of raw SQL
4. Test connectivity: If queries fail, run database test to check connection
5. Be specific: Include relevant WHERE clauses to filter data
6. Protect sensitive data: Don't expose passwords, tokens, or PII unnecessarily

CRITICAL: Always Check Schema Before Writing Queries

⚠️ MANDATORY PROCEDURE: Before writing ANY custom query, you MUST verify the table schema first.

This prevents column name errors and ensures your queries work correctly.

Workflow:

1. Identify the tables you need to query
2. Check the schema of each table using describe
3. Write your query using the actual column names from the schema
4. Execute the query

Example Workflow:

❌ WRONG - Don't do this:

# Directly querying without checking schema
database query "SELECT id, name, status FROM projects_task LIMIT 5"
# Error: Unknown column 'status' - should be 'status_id'

✅ CORRECT - Always do this:

# Step 1: Check the schema first
database describe projects_task

# Step 2: Review the output to see actual columns
# Output shows: id, code, name, status_id, plan_date, etc.

# Step 3: Write query with correct column names
database query "SELECT id, code, name, status_id, plan_date FROM projects_task LIMIT 5"

Real Example with JOINs:

# Step 1: Check both table schemas
database describe sytexauth_user
database describe people_profile

# Step 2: Identify join columns and field names
# sytexauth_user has: id, email, profile_id
# people_profile has: id, name

# Step 3: Write the JOIN query correctly
database query "
SELECT u.id, u.email, p.name
FROM sytexauth_user u
JOIN people_profile p ON u.profile_id = p.id
WHERE u.id = 1
"

Quick Schema Reference Commands:

# List all tables
database tables

# Describe a specific table (shows columns, types, keys)
database describe projects_task

# Describe multiple tables before complex queries
database describe sytexauth_user
database describe people_profile
database describe projects_project

Working with Multiple Databases

When you have multiple databases configured (e.g., production, staging, local), you can easily switch between them:

# List all configured databases
database dbs

# Compare data between environments
database --db production query "SELECT COUNT(*) FROM projects_task"
database --db staging query "SELECT COUNT(*) FROM projects_task"

# Check schema differences
database --db production describe projects_task
database --db staging describe projects_task

# Debug issues by checking production vs staging
database --db production query "SELECT * FROM projects_task WHERE id = 47515"
database --db staging query "SELECT * FROM projects_task WHERE id = 47515"

# Test queries in staging before running in production
database --db staging query "SELECT COUNT(*) FROM users WHERE last_activity > NOW() - INTERVAL 30 DAY"
# If it works, run in production:
database --db production query "SELECT COUNT(*) FROM users WHERE last_activity > NOW() - INTERVAL 30 DAY"

Default Database:
If you don't specify --db, the first configured database is used by default. You can set a default by configuring SYTEXDB_DEFAULT_DB in your .env file.

Query Examples

Find tasks by status:

database query "SELECT id, title, status FROM tasks WHERE status = 'in_progress' LIMIT 10"

Count records by type:

database query "SELECT status, COUNT(*) as count FROM tasks GROUP BY status"

Join queries:

database query "
SELECT t.title, p.name as project_name
FROM tasks t
JOIN projects p ON t.project_id = p.id
LIMIT 10
"

Check specific record:

database query "SELECT * FROM tasks WHERE id = 12345"

Troubleshooting

Connection Errors

If you get connection errors:

1. Run database test to verify credentials
2. Check that database host is accessible
3. Verify database user has SELECT permissions

Query Errors

• Syntax errors: Double-check SQL syntax
• Table doesn't exist: Use database tables to see available tables
• Column doesn't exist: Use database describe <table> to see columns
• Blocked query: Only SELECT queries allowed - modify your query to read-only

Missing mysql Client

If you get "mysql client not found":

• macOS: brew install mysql-client
• Linux: apt-get install mysql-client or yum install mysql

Security Notes

• This skill uses a read-only database user
• All queries are validated to prevent modifications
• Credentials are stored in .env file (never commit this)
• Production data should be treated as confidential
• Don't share query results containing sensitive information

Getting Help

For command usage:

database help

For table-specific information:

database describe <table_name>