Docs

Four ways to use DQaaS. All four share the same checks and report format. Browser results are computed on the device.

1. CLI

A single native command-line program. It profiles and checks files on your own machine with no server and nothing uploaded, and it also works as a client for the API. Platforms: Linux x86_64 and macOS on Apple silicon.

Install the dqaas CLI and the dqaas-mcp server. The script checks the SHA-256 of the download before installing (to ~/.local/bin, or DQAAS_INSTALL_DIR):

curl -fsSL https://dl.dqaas.io/install.sh | sh

The Linux build needs glibc 2.35 or newer (Ubuntu 22.04, Debian 12). Checksums: https://dl.dqaas.io/0.1.0/SHA256SUMS.

dqaas run customers.csv -o report.html

As an API client, it reads DQAAS_URL and DQAAS_API_KEY, or the file written by dqaas login:

export DQAAS_URL=https://api.dqaas.io
dqaas login --token dqat_...
dqaas upload orders.csv --project PROJECT_ID --wait
dqaas analyze DATASET_ID --wait
dqaas rules DATASET_ID --suggest
dqaas report DATASET_ID --format html -o report.html
dqaas search customer
dqaas impact DATASET_ID --verified-only

2. REST API and OpenAPI

Authenticate with Authorization: Bearer dqat_.... Create tokens in the console under Tokens, with read, write or admin scope. The generated OpenAPI 3.1 document is at https://api.dqaas.io/openapi.json.

curl -H "Authorization: Bearer $DQAAS_API_KEY" https://api.dqaas.io/v1/datasets
POST /v1/datasets/upload
POST /v1/datasets/{id}/analyze
GET  /v1/datasets/{id}/rules/suggestions
GET  /v1/datasets/{id}/report?format=json|md|html|pdf
GET  /v1/catalogue/search?q=customer
GET  /v1/datasets/{id}/impact

3. MCP for AI agents

The hosted endpoint is POST https://api.dqaas.io/mcp with Authorization: Bearer dqat_... (the 15 API-backed tools, with your token's permissions; browser origins are refused). The CLI package also includes a local stdio MCP server with 16 tools: dataset listing, profiling, rule suggestion, reports, impact analysis and more. Server tools use the CLI credentials (run dqaas login); profile_file works with no account.

4. In the browser

Sign in, open Attach locally, and choose a CSV, TSV, JSON, JSONL or Parquet file. It is profiled and checked on your own device and is never uploaded. Results are computed locally and are not verified by the server. You can optionally publish only the profile and rule results to the catalogue.