AI Agent Database Workbench — Query Your Agent Data with SQL (2026)

Every session, board issue, note, schedule, and CRM contact amux manages lives in a local SQLite database. The Database tab puts a full SQL workbench on top of it — schema browser, query editor, results table. Updated July 2026.

What Data amux Stores

Every action amux takes — starting a session, creating a board issue, saving a note, sending an email — writes to ~/.amux/amux.db. The database is the single source of truth for your entire AI engineering operation. Key tables:

How to Use the Database Tab

1. Open the amux dashboard at https://localhost:8822 (or the iOS app).
2. Click the Database tab in the top tab bar. You will see a schema browser panel on the left showing all tables and their columns.
3. Browse the schema by clicking any table name — the editor pre-fills a SELECT * FROM <table> LIMIT 50 and runs it immediately.
4. Write your SQL in the query editor. Click Run or press Ctrl+Enter (Cmd+Enter on macOS). Results appear in the table below with column headers, row count, and execution time.
5. Export results by selecting and copying rows from the result table, or use the sqlite3 CLI for bulk CSV export.

10 Power Queries for Your AI Engineering Team

Copy these directly into the Database tab query editor:

1. Which agents ran the most sessions this week?

SELECT session, COUNT(*) AS runs
FROM sessions
WHERE created_at > datetime('now', '-7 days')
GROUP BY session
ORDER BY runs DESC;

2. What tasks have been stuck in "doing" for more than 24 hours?

SELECT title, session, created
FROM board_issues
WHERE status = 'doing'
  AND created < datetime('now', '-1 day')
ORDER BY created ASC;

3. How many sessions per AI provider?

SELECT provider, COUNT(*) AS total
FROM sessions
GROUP BY provider
ORDER BY total DESC;

4. Board items completed per session this month

SELECT session, COUNT(*) AS completed
FROM board_issues
WHERE status = 'done'
  AND updated > datetime('now', 'start of month')
GROUP BY session
ORDER BY completed DESC;

5. All sessions with an error or failed status

SELECT name, status, description, created_at
FROM sessions
WHERE status IN ('error', 'failed', 'crashed')
ORDER BY created_at DESC
LIMIT 20;

6. CRM contacts added this month

SELECT name, company, email, created_at
FROM crm_contacts
WHERE created_at > datetime('now', 'start of month')
ORDER BY created_at DESC;

7. Schedule run frequency — how often does each schedule fire?

SELECT title, schedule_expr, last_run, next_run, enabled
FROM schedules
ORDER BY next_run ASC;

8. Notes created recently

SELECT slug, title, created_at, updated_at
FROM notes
WHERE trashed = 0
ORDER BY updated_at DESC
LIMIT 20;

9. Active sessions right now

SELECT name, status, description, project, provider
FROM sessions
WHERE status = 'active'
ORDER BY created_at DESC;

10. Sessions grouped by project

SELECT project, COUNT(*) AS session_count,
       SUM(CASE WHEN status='active' THEN 1 ELSE 0 END) AS active_now
FROM sessions
WHERE project IS NOT NULL
GROUP BY project
ORDER BY session_count DESC;

Database Workbench vs Alternatives

Backup Your Agent Database

amux.db uses SQLite WAL mode for durability, but a nightly backup is good practice — especially if you are storing CRM relationships, research notes, and board history built up over months. Use the sqlite3 CLI backup command:

# One-time backup
sqlite3 ~/.amux/amux.db ".backup ~/.amux/amux-backup-$(date +%Y%m%d).db"

# Schedule via amux (runs nightly at 2am on any registered session)
curl -sk -X POST https://localhost:8822/api/schedules \
  -H 'Content-Type: application/json' \
  -d '{
    "title": "Nightly amux.db backup",
    "session": "your-session",
    "command": "sqlite3 ~/.amux/amux.db \".backup ~/.amux/amux-backup-$(date +%Y%m%d).db\"",
    "schedule_expr": "daily at 2am"
  }'

Frequently Asked Questions

Beyond the Database — What the API Gives You

The amux REST API exposes all the same data with richer filtering, pagination, and write support. For quick ad-hoc exploration, the Database tab is fastest. For building integrations (n8n workflows, scripts, dashboards), the API is the right layer:

# Board issues — filtered, paginated, JSON
curl -sk https://localhost:8822/api/board?status=done&since=2026-07-01

# Sessions list
curl -sk https://localhost:8822/api/sessions

# CRM contacts
curl -sk https://localhost:8822/api/crm/contacts

# Schedules
curl -sk https://localhost:8822/api/schedules

git clone https://github.com/mixpeek/amux && cd amux && ./install.sh
amux register myproject --dir ~/Dev/myproject --yolo
amux start myproject
# Open https://localhost:8822 → Database tab

Related: Agent coordination · Agent fleet operations · Scheduling AI agents · All guides →