dart-query

MCP server for Dart AI task management, optimized for batch operations and minimal context usage.

Instead of looping through tasks one-by-one (filling your context window with intermediate JSON), dart-query uses DartQL selectors and server-side batch operations to update hundreds of tasks in a single call. A 50-task update that would normally consume ~30K tokens takes ~200 tokens with zero context rot.

Quick Start

1. Get Your Dart AI Token

Visit https://app.dartai.com/?settings=account and copy your token (starts with dsa_).

2. Configure MCP

npx (recommended)

{
  "mcpServers": {
    "dart-query": {
      "command": "npx",
      "args": ["-y", "@standardbeagle/dart-query"],
      "env": {
        "DART_TOKEN": "dsa_your_token_here"
      }
    }
  }
}

SLOP-MCP (v0.10.0+)

slop register dart-query \
  --command npx \
  --args "-y" "@standardbeagle/dart-query" \
  --env DART_TOKEN=dsa_your_token_here \
  --scope user

3. Verify

info({ level: "overview" })

4. Example: Batch Update

// Preview first
batch_update_tasks({
  selector: "dartboard = 'Engineering' AND priority = 'high'",
  updates: { status: "Doing" },
  dry_run: true
})

// Execute
batch_update_tasks({
  selector: "dartboard = 'Engineering' AND priority = 'high'",
  updates: { status: "Doing" },
  dry_run: false
})

Tools

See TOOLS.md for full parameter references, DartQL syntax, and CSV import format.

DartQL Selectors

SQL-like WHERE clauses for targeting tasks in batch operations:

dartboard = 'Engineering' AND priority = 'high' AND tags CONTAINS 'bug'
due_at < '2026-01-18' AND status != 'Done'
title LIKE '%auth%'

Safety

All Dart AI operations are production (no sandbox). dart-query provides:

• Dry-run mode on all batch operations — preview before executing

• Validation phase for CSV imports — catch errors before creating anything

• Confirmation flag (confirm: true) required for batch deletes

• Recoverable deletes — tasks move to trash, not permanent deletion

License

MIT