SQL Interface

Query your agent's database directly with the SQL editor.

Overview

Features:

• Monaco editor with SQL syntax highlighting
• Schema browser
• Query execution with timing
• Result display and export

Using the Editor

Via Dashboard

1. Navigate to SQL
2. Write your query
3. Click Execute or press Ctrl+Enter
4. View results

The execute_sql tool

The agent's own execute_sql tool is narrower than the editor: read-only, one statement, and tenant-scoped. Only an allowlist of tables with an agent_id column can be named (contacts, conversations, chat_messages, calendar_tasks, schedules, transactions, workflows, agent_tasks, turn_traces, jev_decisions, data_sources, media, skills, token_usage, …), and every referenced table is shadowed by a CTE filtered to the agent, so a query cannot read another agent's rows however it is written. Platform tables (agents, tenants, keys, sessions) are not reachable from it; schema-qualified names and multiple statements are rejected.

Allowed Operations (dashboard editor)

Operation	Allowed
SELECT	Yes
INSERT	Yes
UPDATE	Yes
DELETE	Yes
PRAGMA	Yes (read-only)
EXPLAIN	Yes
CREATE	No
DROP	No
ALTER	No

Schema Browser

View database structure:

• Tables
• Columns
• Types
• Indexes

Query Examples

List Workflows

SELECT id, name, status, created_at
FROM workflows
WHERE agent_id = 'current-agent'
ORDER BY created_at DESC
LIMIT 10

Count Contacts

SELECT COUNT(*) as total
FROM contacts
WHERE agent_id = 'current-agent'

Search Messages

SELECT content, created_at
FROM chat_messages
WHERE content LIKE '%keyword%'
ORDER BY created_at DESC

Best Practices

1. Use Indexes

Query on indexed columns for performance.

2. Limit Results

Always use LIMIT for large tables:

SELECT * FROM logs LIMIT 100

3. Select Needed Columns

-- Good
SELECT name, email FROM contacts

-- Avoid
SELECT * FROM contacts

API Reference

See SQL API for complete endpoint documentation.