Build with the QUESTIONSTAR API

Read survey structures, submit and retrieve responses, push results to your systems with webhooks, export files and run server-side statistics — over a clean REST API. AI agents connect through our MCP server.

Full API reference Open the app

API keys live in the app: user menu → Integrations → API Keys → Generate API key.

Quick start

Three steps from zero to your survey data.

Create an API key

In the QUESTIONSTAR app: user menu → Integrations → API Keys → Generate API key. Every request carries this token in the Authorization header — the raw value, no Bearer prefix.

List your surveys

curl https://api.questionstar.de/api/v1/Survey/Surveys \
  -H "Authorization: YOUR_API_TOKEN"
# → every survey with its 8-character code — the code addresses the survey in all other calls

Read the responses

curl -X PUT https://api.questionstar.de/api/v1/ResultCase/ResultCases/Survey/Cases/1a2b3c4d \
  -H "Authorization: YOUR_API_TOKEN" -H "Content-Type: application/json" \
  -d '{"IncludeRespondents": false}'
# → questions as columns (Q1, Q2.3 …), one row per response

Every JSON answer arrives in a common envelope: Success / Data / Errors. Business errors come as HTTP 200 with Success: false.

Recipes

The three integrations almost everyone builds first.

Push responses to your system

Create a webhook and get an HTTP call the moment a response arrives — no polling. Test it with one request, inspect the delivery log, subscribe to SurveyPublished for structure syncs.

POST /api/v1/WebHook/WebHooks/Survey/{code}

Export result files

Excel, CSV or SPSS — one request, one file, however many rows. 14,000 responses export in seconds.

GET /api/v1/Export/Exports/Survey/Cases/XLSX/{code}

Incremental sync for BI

Pull only what changed since your last run with ChangedSince, filter by device or traffic source. Use an overlapping window and de-duplicate by response ID.

…/Survey/Cases/{code} + filter

Rate limits

Counted per token and per request — a 10,000-row export is one request.

Class	Default	Applies to
light	120 / minute	most operations
heavy	600 / hour	tabular response reads, statistics
export	12 / hour	file exports

On limit you receive HTTP 429 with Retry-After — wait it out instead of retrying in a loop. Full details in the API reference introduction.

Beta

MCP server for AI agents

Claude and other AI agents work with your surveys through our Model Context Protocol server: 21 tools — from reading results and file exports to cloning surveys, publishing and server-computed statistics (correlation, t-test, regression — the server calculates, the agent narrates).

Currently in closed beta. Sign-in with your QUESTIONSTAR account is coming; contact us to try it early. More about the MCP integration →

Coming soon

SDKs

Official TypeScript and Python clients, generated from the live API specification — with the full endpoint documentation in your editor's autocomplete.