Dataify DuckDuckGo Search

Workflow

Use python3 to run the bundled script for the entire flow. Do not build the HTTP request manually unless the script needs maintenance.

Always preview parameters before every API call:

python3 scripts/duckduckgo_search.py --request "\x3Cuser request>" --preview

On Windows workspaces where the python3 alias is unavailable, use the installed Python 3 launcher for the same script, for example python scripts/duckduckgo_search.py ....

Show the preview table to the user. The table must include the complete field list except Authorization, and only these columns: parameter name, current value, default value, and description. Ask whether the user wants to modify any parameter. Do not call the API until the user confirms.

After the user confirms, call the API with the same request and explicit overrides, adding --confirmed. If the user provides a token in the conversation, pass it explicitly:

python3 scripts/duckduckgo_search.py --request "\x3Cuser request>" --token "\x3CDATAIFY_API_TOKEN>" --confirmed

The script reads DATAIFY_API_TOKEN from the environment when --token is not provided. If no token is available, stop and ask the user to provide a Dataify API token or register at Dataify Dashboard.

Field Mapping

Pass the user's full request to --request; the script automatically maps natural-language hints and explicit assignments to Dataify fields:

Use default values from parameter descriptions when the user does not specify a field: engine=duckduckgo, json=1, search_assist=false, safe=-1, start=0, m=50, and no_cache=false. q, kl, and df have no default. Do not treat API documentation examples as defaults: never use q=pizza, kl=us-en, search_assist=true, safe=1, df=d, start=0 because it appeared in an example, m=10, or no_cache=true unless the user request or the documented default says so.

For exact control, pass explicit flags such as --q, --json, --kl, --safe, --df, --start, --m, --no-cache, and --search-assist; explicit flags override the natural-language parser.

Response Handling

The script submits the request as application/x-www-form-urlencoded form data, not JSON.

Return the script stdout directly to the user. Do not summarize, translate, pretty-print, filter, or otherwise process the API response.