Dory MCP Server

AI agents can now write SQL, inspect schemas, and answer analytical questions. But raw agent output is not enough for real data work.

- see the exact SQL that was run
- inspect real result sets
- fix wrong queries
- turn results into charts or exports
- preserve context across multiple steps
- continue the work in a real SQL workspace

Dory provides the same underlying database actions to both the UI and agents:

- Humans use Dory as a SQL client: write SQL, browse schema, run queries, filter results, create charts, and save queries.
- Agents use Dory through MCP: list connections, explore schemas, run read-only SQL, create tabs, and organize database work.
- Agent-generated work becomes a real workspace that humans can review, edit, and continue.

Agent database work should not disappear into a chat transcript.

- Open agent-created SQL tabs as normal workspace tabs
- Inspect SQL, result sets, filters, charts, and saved context
- Edit generated SQL and rerun it yourself
- Continue an agent run from the same workspace instead of restarting from scratch

@getdory/cliis the agent entrypoint for running Dory without opening the desktop app.

- Works with Claude Code, Codex CLI, and other MCP-compatible clients
- Runs standalone MCP servers over stdio or HTTP
- Acts as a headless runtime for automation and direct Dory Actions
- Can use standalone data, hosted Dory Web, or Desktop local data with--data desktop

- Multi-tab SQL editor with multiple result sets
- Schema browser for tables, columns, and database objects
- Saved queries for reusable analysis
- Built-in query execution history and workspace context

Understand unfamiliar databases at a glance with an interactive map of tables and relationships.

- Visualize primary keys, foreign keys, and relationships across schemas
- Search for tables, select schemas, and switch between all columns and keys only
- Auto-arrange the graph, fit it to the viewport, and open any table directly
- Export the complete schema graph as PNG or SVG

- Inspect real result sets in a table view
- Filter, search, and review returned rows
- Turn query output into charts directly inside the workspace
- Keep results attached to the SQL that produced them

An AI assistant grounded in real database schema and current query context.

- Generate SQL from natural language
- Rewrite, fix, and explain SQL in the current tab
- Use current database schema and query context
- Keep AI assistance inside the real SQL workspace

- Save useful SQL as reusable queries
- Organize query work across connections and workspaces
- Let humans and agents build on previous database work
- Preserve context beyond a single chat response

Dory is a multi-database SQL client, with broad driver support and deeper integrations where Dory can provide more than generic SQL execution.

Dory includes native ClickHouse operations surfaces for teams that need more than a generic SQL editor.

- Query monitoring with slow queries, errors, active users, latency, and throughput
- Multi-dimensional filtering by user, database, query type, and time range
- User and role management without hand-writing everyGRANTorCREATE USERstatement
- Cluster-level privilege operations with On Cluster support

Make sure Docker is installed, then run:

docker run -d --name dory \ -p 3000:3000 \ -e DS_SECRET_KEY="$(openssl rand -base64 32 | tr -d '\n')" \ -e BETTER_AUTH_SECRET="$(openssl rand -hex 32)" \ -e BETTER_AUTH_URL="http://localhost:3000" \ -e DORY_AI_PROVIDER=openai \ -e DORY_AI_MODEL=gpt-4o-mini \ -e DORY_AI_API_KEY=your_api_key_here \ -e DORY_AI_URL=https://api.openai.com/v1 \ -e NEXT_PUBLIC_REQUIRE_EMAIL_VERIFICATION=false \ -e DORY_INIT_USER_EMAIL=admin@getdory.dev \ -e DORY_INIT_USER_PASSWORD=admin \ dorylab/dory:latest

The initial administrator account is controlled byDORY_INIT_USER_EMAILandDORY_INIT_USER_PASSWORDin.env.

For long-running self-hosted deployments, Docker Compose runs Dory with a dedicated PostgreSQL database and persistent volumes.

cp docker-compose.env.example .env # Edit .env and replace all placeholder secrets/passwords. docker compose up -d

Install the Dory skill for Codex, ChatGPT, Claude, and other agents that support the open Agent Skills format.

npx skills add https://github.com/dorylab/skills --skill dory

The skill teaches agents when to use Dory, how to use Dory MCP tools, and how to preserve generated SQL, result sets, and findings in an editable Dory Agent Run workspace.

The skill does not install the Dory CLI or configure MCP automatically. For live database access, connect Dory MCP with the setup commands below.

Use@getdory/clito connect MCP-compatible agents to Dory. It can run a local stdio MCP server, host a long-running HTTP MCP endpoint, bridge to hosted Dory Web, or reuse Dory Desktop local data with--data desktop.

Run a local stdio MCP server when your agent client is on the same machine:

codex mcp add dory -- npx -y @getdory/cli mcp serve --stdio --data standalone codex mcp list

claude mcp add dory -- npx -y @getdory/cli mcp serve --stdio --data standalone claude mcp list

Run a local HTTP MCP endpoint when you want a long-running service:

npx -y @getdory/cli mcp token create --data standalone --name "local-http" export DORY_MCP_TOKEN="dory_mcp_..." npx -y @getdory/cli mcp serve \ --http \ --host 127.0.0.1 \ --port 3318 \ --token "$DORY_MCP_TOKEN" \ --data standalone

codex mcp add \ --url http://127.0.0.1:3318/api/mcp \ --bearer-token-env-var DORY_MCP_TOKEN \ dory

claude mcp add \ --transport http \ dory \ http://127.0.0.1:3318/api/mcp \ --header "Authorization: Bearer $DORY_MCP_TOKEN"

Bridge hosted Dory Web to local MCP clients:

npx -y @getdory/cli mcp login --url https://your-dory-host codex mcp add dory-hosted -- npx -y @getdory/cli mcp bridge --url https://your-dory-host claude mcp add dory-hosted -- npx -y @getdory/cli mcp bridge --url https://your-dory-host

Dory is built with a pluggable AI provider architecture. You can freely switch between different model vendors by changing environment variables, with no code changes required.

- Next.js + React + Tailwind
- Drizzle ORM
- Multi-model AI SDK integration
- PGLite
- Resend
- Shadcn UI
- Monaco Editor

- Data engineers
- Data analysts
- Database platform teams
- Agent builders who need a database execution layer
- ClickHouse operations teams

Your data stays yours. Except for AI requests sent through Cloudflare Gateway, everything — connections, tabs, results, and saved queries — is stored locally on your device.

Official Airtable MCP server and skills for working with bases, records, workflows, and business operations from AI agents.

MCP Server For Apache Doris, an MPP-based real-time data warehouse.

Official MCP Server from Atlan which enables you to bring the power of metadata to your AI tools

Query Onchain data, like ERC20 tokens, transaction history, smart contract state.

Read and write access to your Baserow tables.

Introspect and query your apps deployed to Convex.

Interact with the data stored in Couchbase clusters using natural language.