opteryx-mcp
OfficialClick on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@opteryx-mcpfind datasets about customers and show their schemas"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
opteryx-mcp
Connect Claude Desktop, Claude Code and other MCP clients to Opteryx. Ask questions about your data in plain English; the assistant finds datasets, reads schemas, queries, and checks Opteryx SQL, all as you and against only the data you can see.
uvx opteryx-mcp login --user <user>Why a local bridge
The Opteryx MCP endpoint (https://agent.opteryx.app/mcp/) takes a bearer
token, and access tokens last five minutes. A token pasted into a client's
config stops working almost immediately. opteryx-mcp holds a user and a
personal access token instead, exchanges them for a fresh access token a
minute before each one expires, and forwards everything the client sends.
It also re-mints if a token is rejected, and re-opens the MCP session if the
server forgets it.
It has no dependencies beyond the Python standard library.
Related MCP server: Syntitan MCP Server
Setup
1. Create a personal access token
In Opteryx Studio, open Settings and create a personal access token. It
looks like opt_..._01, lasts 90 days by default, and is shown once.
2. Store it
On macOS, login stores the token in the Keychain and checks it works.
security prompts for the token, so it never appears in shell history:
uvx opteryx-mcp login --user <user>Connected to https://agent.opteryx.app/mcp/ as <user>.
7 tools: search_datasets, get_dataset_schema, query_dataset, profile_column, lookup_sql_syntax, search_docs, validate_sqlRun it again to replace an expired token. On other platforms, set
OPTERYX_TOKEN in the client's config instead.
3. Add it to your client
Claude Desktop. Print the config entry:
uvx opteryx-mcp config --user <user>Then quit Claude Desktop (Cmd-Q) and merge the entry into
~/Library/Application Support/Claude/claude_desktop_config.json, keeping the
file's other keys. Desktop rewrites that file when it quits, so an entry added
while it is running is silently dropped. The entry looks like:
{
"mcpServers": {
"opteryx": {
"command": "/opt/homebrew/bin/uvx",
"args": ["opteryx-mcp", "--user", "<user>"]
}
}
}uvx is named by absolute path because Desktop starts servers with a minimal
PATH. Reopen Desktop and the server appears in a chat's + / tools menu,
under Settings → Developer, and in Code-tab sessions.
Claude Code.
claude mcp add opteryx -- uvx opteryx-mcp --user <user>Other clients. Any client that launches stdio servers takes the same
command: uvx opteryx-mcp --user <user>.
Tools
Tool | What it does |
| Find datasets by name or column name |
| Columns, types, statistics and sample rows |
| Run an OData v4 query; reads are capped at 200 rows |
| The values a column actually holds, before you filter on one |
| Confirm a function or statement exists in Opteryx SQL |
| Check SQL against the dialect and your schemas without running it |
| Search the Opteryx documentation |
Every call runs as the user the token belongs to.
Configuration
Variable | Default | Purpose |
| — | The user, if |
| Keychain entry | The personal access token |
|
| MCP endpoint |
|
| Where access tokens are minted |
Troubleshooting
uvx opteryx-mcp check --user <user>connects once and lists the tools, the same way a client would.Claude Desktop writes the bridge's log to
~/Library/Logs/Claude/mcp-server-opteryx.log. It records each token minted and every request that fails.token request failed (401)means the user and token don't match, or the token has expired or been revoked. Create a new one and runloginagain.If the server never appears in Desktop, check that the entry is still in the config file. If you edited it while Desktop was running, it was overwritten.
Development
pip install -e . pytest ruff
make test
make lintReleases are published to PyPI from v* tags.
This server cannot be deployed
Maintenance
Related MCP Connectors
Query your warehouse or a CSV with Claude/ChatGPT over MCP, governed by table-level ACL + audit.
Query your org's data in natural language — read-only MCP access to SQL, NoSQL, files & warehouses.
Governed data discovery, exact queries, decisions, simulations, and runtime utilities over MCP.
Query, join, profile, clean and convert CSV/JSON/Parquet with server-side DuckDB over MCP.
Related MCP Servers
- FlicenseAqualityDmaintenanceEnables querying, exploring, and interacting with Azure SQL databases through MCP-compatible clients like Claude Desktop.5-

Syntitan MCP Serverofficial
FlicenseNot gradedqualityDmaintenanceLets Claude and other MCP-compatible clients securely connect to Syntitan, an AI-Ready Data Platform, through natural language for data management and analytics.-- FlicenseNot gradedqualityCmaintenanceEnables local MCP-compatible clients, such as Claude Desktop, to query an AWS Athena database with read-only tools for listing tables, describing schemas, running SELECT queries, sampling rows, and getting row counts.-
- AlicenseNot gradedqualityCmaintenanceEnables MCP clients such as Claude, ChatGPT and Cursor to answer business questions about sales, customers, orders and spreadsheets in natural language by reading live Postgres/SQLite databases and Google Sheets without exporting files or gaining write access. Guardrails include AST-based read-only SQL validation, read-only database roles, row/byte caps with pagination, PII masking, bearer-token auth for remote HTTP use, and a full JSONL audit log of every tool call.MIT