ATX AnalyticsTable API Reference v1 | Every Endpoint, OpenAPI 3.1

API reference — v1Every endpoint, grouped by tag, generated from the OpenAPI 3.1 document the product publishes. This reference is rendered from the same OpenAPI 3.1 document an installation serves at GET /api/v1/openapi.json. Paths are relative to your installation's URL. Every call needs a credential — a scoped service token as Authorization: Bearer <token> is recommended — and every error is an RFC 9457 problem document with a stable code.

Discovery

GET /api/v1 — API root. Version, build and links to every resource collection. Start here to check connectivity and credentials. Role: viewer.
GET /api/v1/openapi.json — OpenAPI 3.1 document. This API described in OpenAPI 3.1 — import it into Postman or an API gateway, or generate a typed client. Role: viewer.
GET /api/v1/docs — Reference page. A human-readable reference of this API, generated from the same catalogue as the OpenAPI document. No scripts; works offline. Role: viewer.
GET /api/v1/me — Who am I. The identity this credential resolves to, its role and permissions, and the rate limit that applies to it. Role: viewer.
GET /api/v1/health — Service health. Readiness of the query engine and of every registered MCP server (circuit-breaker state, in-flight calls). status is ok, degraded or draining. Role: viewer.

Data

GET /api/v1/tables — List tables. Tables and views of the project's database that this credential may see (row and column security apply). Role: viewer.
GET /api/v1/tables/{name} — Describe a table. Columns and types of one table, after column-level security. Role: viewer.
POST /api/v1/query — Run a read-only SQL query. Runs ONE read-only DuckDB statement on the project database, through the same admission lanes, time limits, row/column security and audit trail as the desktop app. Results page with an opaque cursor. Role: viewer.
GET /api/v1/connections — List saved connections. Saved database connections this credential may use — id, name, kind and database only. Never a host, user or secret. Role: viewer.
POST /api/v1/connections/{id}/query — Run a read-only query on a saved connection. ONE SELECT/WITH statement in the connection's dialect (PostgreSQL, SQL Server, BigQuery, Snowflake, Databricks, Oracle, …), with the stored credential. Object-level access applies. Role: viewer.

Metrics

GET /api/v1/metrics — Metric catalogue. The governed semantic layer: measures, calculated members, dimensions, filters and parameters. Query any measure by any compatible dimension with POST /api/v1/metrics/query — the same definitions the dashboards use. Role: viewer.
GET /api/v1/metrics/{id} — One metric. A measure, calculated member, dimension, filter or parameter by URN or id. For a measure, also the dimensions it can be broken down by. Role: viewer.
POST /api/v1/metrics/query — Headless metric query. Ask for measures by dimensions with filters and parameters; the semantic planner writes the SQL (joins through declared joins only, dialect-correct literals, fan-out warnings in notes), runs it on the dataset's connection and returns rows. dryRun: true returns the plan only. Object-level access to the connection applies. Role: viewer.

Dashboards

GET /api/v1/dashboards — List semantic dashboards. Dashboards this credential may open. Role: viewer.
GET /api/v1/dashboards/{urn} — One dashboard. The full definition: widgets, layout, filters, parameters. Role: viewer.
POST /api/v1/dashboards/{urn}/widgets/{wid}/data — Data behind one widget. Plans and runs one widget exactly as the dashboard would. Body: { parameters?, activeFilterIds? }. Role: viewer.

MCP

GET /api/v1/mcp/servers — List MCP servers. The Model Context Protocol servers registered on this installation (BigQuery, PostgreSQL, Jira, Confluence, …) and their health. No secret value is ever returned. Role: viewer.
GET /api/v1/mcp/templates — List MCP server templates. Ready-made definitions an administrator can start from (BigQuery, PostgreSQL, Jira and Confluence Cloud and Data Center, custom), each with its parameters: name, where it goes, whether it is required, default, example and meaning. Role: viewer.
GET /api/v1/mcp/servers/{id}/tools — List the tools of one MCP server. Every tool the server offers, its classification (read, sql, write) and whether this installation's policy allows calling it. Role: viewer.
POST /api/v1/mcp/servers/{id}/tools/{tool}/call — Call an MCP tool. Calls one allowed tool with { "arguments": { … } }. SQL arguments are checked to be read-only first; write tools are refused on read-only servers. Protected by a per-server deadline, circuit breaker and concurrency limit. Send an Idempotency-Key to make retries safe. Role: analyst.

Error codes

bad_request — HTTP 400, not retryable: The request is not valid
unauthorized — HTTP 401, not retryable: Authentication required
forbidden — HTTP 403, not retryable: Not allowed for this credential
not_found — HTTP 404, not retryable: Not found
method_not_allowed — HTTP 405, not retryable: Method not allowed
conflict — HTTP 409, retryable: Conflict
idempotency_mismatch — HTTP 422, not retryable: Idempotency key reused with a different request
read_only_sql — HTTP 422, not retryable: Only read-only SQL is accepted
planner_error — HTTP 422, not retryable: The metric query cannot be planned
query_error — HTTP 422, not retryable: The database rejected the query
payload_too_large — HTTP 413, not retryable: The result is too large
rate_limited — HTTP 429, retryable: Too many requests
client_closed — HTTP 499, retryable: The client closed the request
internal — HTTP 500, retryable: Internal error
upstream_error — HTTP 502, retryable: An upstream service failed
unavailable — HTTP 503, retryable: Temporarily unavailable
upstream_timeout — HTTP 504, retryable: An upstream service timed out
disabled — HTTP 503, not retryable: This feature is turned off on this installation

Glossary terms on this page

KPI dashboard: definition

ATX Institute on LinkedIn

This site needs JavaScript for the full experience. Every page above is reachable without it.