Simple. Fast. Cost saving. A database client that fits your workflow.
Local-first. Multi-database. No limits. No Extra Cost.
Beta — Under active development. Frequent updates, occasional rough edges. Report issues here.
I built Query Pilot because I couldn't find a database client that worked the way I wanted. Most tools out there are either expensive, way too limited on free tiers, or just missing the things I actually need when debugging production databases day to day.
No existing database client deeply integrates Claude Code or Codex CLI either — so Query Pilot uses them as built-in AI agent runtimes, alongside bring-your-own-key support for other providers.
Source code is currently closed but may be open-sourced in the future.
- No extra AI costs — Uses Claude Code and Codex CLI as AI runtimes directly inside the app. No separate API bills — if you already have Claude Code or Codex, your AI-powered database workflows are covered.
- Multi-connection workspaces — Open multiple database connections side by side in a single workspace. Compare data across PostgreSQL, MySQL, MongoDB, and more without switching windows.
- 100% local — Your data never leaves your machine. Credentials stored in the OS keychain. No cloud. Telemetry off by default.
- Native speed — Rust backend with sub-second startup. Handles 100k+ row tables without breaking a sweat.
| SQL | Document | Key-Value |
|---|---|---|
| PostgreSQL | MongoDB (beta) | Redis |
| MySQL | ||
| MariaDB | ||
| SQLite | ||
| SQL Server |
Create named workspaces with multiple database connections open simultaneously. Each workspace persists its layout, tabs, and state across sessions. Open separate windows per workspace, each with independent panels and editors.
Use Claude Code and Codex CLI directly from the app as AI agent runtimes — no separate API keys or usage fees required. Generate SQL from natural language, get query explanations, explore schemas conversationally, and execute AI-suggested queries with permission gating.
Also supports bring-your-own-key with OpenAI, Anthropic, Google, Mistral, and local models via Ollama. Configure temperature, system prompts, and model selection per provider. Attach table context and images to conversations.
Inline cell editing with full undo/redo. Multi-cell selection with aggregate statistics (sum, avg, min, max, unique count). Fill down/right for bulk edits. Column-level regex filtering. Edit tables even without primary keys. Type-aware cell renderers for JSON, dates, UUIDs, geometry, enums, booleans, and more.
Pin multiple result grids side by side — compare query outputs, validate JOINs, or monitor tables without switching tabs.
Auto-generated entity-relationship diagrams with foreign key connections. Export to DBML or Mermaid format. Pan, zoom, and interactively explore relationships across your schema.
Define virtual FK relationships on tables that lack formal constraints. Referenced values from related tables are embedded directly into the grid as extra columns via automatic LEFT JOIN queries — see the actual data behind foreign key IDs, not just the IDs themselves. Invaluable for debugging production data.
CodeMirror 6 with schema-aware autocomplete, real-time SQL linting, syntax highlighting, and query formatting. Multi-dialect support (PostgreSQL, MySQL, T-SQL, MongoDB). Unlimited tabs — each with its own editor and result grid. Extract CTE dialog, find & replace, and customizable keybindings.
Interactive query plan trees with cost breakdowns, row estimates vs actuals, and buffer statistics. Multi-dialect EXPLAIN parsing. Spot performance bottlenecks at a glance.
Three-tab inspector panel for any selected cell or row. Tree view with expandable nodes, inline editing, and search. Raw view with pretty-printed JSON. Diff view for character-level comparison across multiple selected rows.
Browse structure, indexes, triggers, partitions, and column definitions from a single tabbed interface. View row counts, table sizes, and column metadata. Edit constraints, defaults, and comments inline.
Visual CREATE TABLE builder — configure columns, types, nullable flags, defaults, constraints, primary keys, and foreign keys without writing SQL. MongoDB collection designer with schema validation, time-series, and capped collection options.
Connect through bastion hosts with key-based auth, SSH agent forwarding, and automatic ~/.ssh/config resolution. Rate-limited port allocation for production network setups.
Four protection levels per connection — from strict read-only to full access — enforced at the backend. Prevent accidental DROP or DELETE on production.
Export query results or entire tables to CSV, JSON, SQL INSERT, and Markdown. Configurable delimiters, batch INSERT generation, and timestamped filenames. Export schema definitions as DDL scripts.
Persistent, searchable query history across all connections. Save frequently-used queries to a library for instant re-execution. Access everything from the command palette (Cmd+K).
Planning time, execution time, total time, and plan buffer hits displayed for every query. Built-in performance awareness for every statement you run.
Dedicated query panels for MongoDB (aggregation pipelines, BSON types) and Redis (all data structures, key browsing). Native experiences — not SQL wrappers.
Light, dark, and system-following themes. Adjustable UI zoom (80–150%). Modern UI built with shadcn/ui and Tailwind CSS.
Credentials stored in the OS keychain. Encrypted vault for API keys. SSH key management. No data sent to external servers. Optional error reporting (Sentry) with no SQL, credentials, or user data included.
Download the latest .dmg from Releases. Supports both Intel and Apple Silicon.
- Open the
.dmgand drag Query Pilot to Applications - On first launch, right-click the app and select Open if macOS shows a security warning
Windows and Linux — Coming soon.