Other

Socrata: DataSF (San Francisco open data) on looot

Socrata: DataSF (San Francisco open data) gives your agent 3 jobs through looot: search open data dataset, query open data dataset, get open data dataset. 3 endpoints run today, some of them free.

For humansFor agents

Category: Other
Access: No key needed: these endpoints are open, looot calls them for you
Price: some of them free
Jobs: 3 jobs, 3 endpoints

Choose Socrata: DataSF (San Francisco open data) when

You want the lowest price to search open data dataset: free, the cheapest of 3 providers on looot.
You want the lowest price to query open data dataset: free, the cheapest of 3 providers on looot.
You want the lowest price to get open data dataset: free, the cheapest of 3 providers on looot.
You want to start without any key: some of its endpoints are open and looot calls them for you.

Endpoints and jobs

Each endpoint runs through looot today. Where a job has a benchmark page, its link compares every provider on that job.

DataSF (San Francisco open data) / Find datasets on data.sf.gov
open-data.dataset.search
free
not enough runs to rank
DataSF (San Francisco open data) / Query a dataset's rows with SoQL
open-data.dataset.query
free
not enough runs to rank
DataSF (San Francisco open data) / Read one dataset's columns and metadata
open-data.dataset.get
free
not enough runs to rank

A minimal request

For DataSF (San Francisco open data) / Find datasets on data.sf.gov (socrata-sf-datasets), free. Replace each <placeholder> with your value and send a new idempotency key per run.

MCP

Call the looot MCP tool "run" with:
{
  "endpointId": "socrata-sf-datasets",
  "input": {},
  "idempotencyKey": "<a new id per run>"
}

REST

curl -X POST https://api.looot.ai/v1/runs \
  -H "Authorization: Bearer $LOOOT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"endpointId":"socrata-sf-datasets","input":{},"idempotencyKey":"<a new id per run>"}'

CLI

looot run socrata-sf-datasets --input '{}'

Response shape

looot has no stored output format for this endpoint yet, so no response shape is shown here. Run it once and read the answer.

How we measure

Run numbers are looot's own runs of each endpoint over the last 30 days: the share that completed without an error, and the median time to answer. An endpoint with fewer than 10 runs in that window shows "not enough runs to rank" and is never ranked. Prices are what looot charges per call or per result, fee included, read from the live catalog; you prepay credit and each call is charged when it runs.

Read 2026-11-10 23:06 UTC.

Give your agent every data provider, one key.

Start your workspace, top up, and paste one prompt into the agent you already use.

Start your workspace See what agents can do