connector-skill

name	connector-skill
description	Connect to data sources, explore schemas, write queries, and profile data quality. Use when the user wants to query a database, explore a data source, profile tables, or understand their data schema. Handles SQL dialects, API-based data sources, and file-based data. Includes query optimization and result validation patterns.

Data Connector Skill

Query Patterns

SQL-based Sources

Write dialect-appropriate SQL. Check schema first, validate results after.

API-based Sources

Use MCP tools to query. Handle pagination and rate limits.

File-based Sources

Read CSV/Excel/JSON files and analyze with pandas or similar.

Validation Checklist

• Row count sanity
• Null check
• Value range check
• Aggregation logic

Trace Awareness

When previous TRACE.md or FILTER.md files exist for this data source:

1. Read the Attention Log — know what schema areas were already explored
2. Read the Confidence Map — know where schema understanding is weak
3. Read the Surplus Value — know what data patterns emerged last time
4. Read the Filter — skip known-noise areas, prioritize signal areas

References

• See references/schema-guide.md for schema exploration patterns