FCC ECFS Filings Watch

Watch FCC Electronic Comment Filing System filings by docket and by filer name. Each later run returns filings whose date_received is at or after the stored watermark and whose ID is not already stored. The first run is a free baseline and stores those watermark filings as uncharged snapshot rows.

This version reads the ECFS filings API only. It does not cover equipment authorization.

What you get

A dataset row for each new filing, plus free baseline rows on the first run for a Monitor ID. One key-value record named OUTPUT holds the whole batch (new filings, baseline snapshots, error rows, and whether the cap stopped the run). News Scout should read OUTPUT when the run finishes. This Actor does not send a webhook per filing.

Field	Meaning
submissionId	ECFS id_submission
dateReceived	FCC date_received
filers, proceedings	Names and docket numbers on the filing
submissionType, filingStatus	ECFS type and status
documentUrls	Document links from the record
textExcerpt	First 500 characters of the filing text
sourceUrl	ECFS search link for the filing
rowType	baseline (free), filing (charged), error (free), or capped (free)

How to use it

1. Add up to 10 dockets and 10 filer names (you can lower that with maxDockets and maxFilers; the hard caps are 25). The status message names the cap.
2. Paste your free api.data.gov key into api.data.gov key. If you leave it empty, or type the public DEMO_KEY, the run uses that shared key. The log and the status message both warn that DEMO_KEY allows about 10 requests, which is enough for only a few dockets.
3. Run once. The dataset gets one free baseline row for each filing at the newest date_received. lastSeen and those filing IDs are stored for this Monitor ID. That baseline is not charged.
4. Schedule the same Monitor ID. The next run stores a row for a filing received at or after lastSeen when that filing's ID is not already stored.

How much does it cost?

Pay per new filing: filing at $0.002. The price is configured in .actor/pricing.json and is not applied on the platform until you say so.

The first baseline is free, including its snapshot rows. An error row is free. A filing past maxFilings is not stored and is not charged; the run adds one free capped row instead. Runs use 128 MB.

Good to know

• API. GET https://publicapi.fcc.gov/ecfs/filings with proceedings.name and filers.name. The key is a query parameter and is not copied into the dataset or the watch state.
• Date filter. The live API accepts date_received=[gte]YYYY-M-D[lte]YYYY-M-D. A timestamp, or [gt] without [lte], returns HTTP 400. [gt] on a calendar day still includes later timestamps that day, and [lte] on a day excludes later timestamps that day. The Actor sends a day window from the lastSeen day through two days ahead of today, then keeps a row when its timestamp is >= lastSeen and its ID is not already stored.
• State. The named store fcc-filings-watch-state keeps seen-<monitorId> with two fields: lastSeen and ids. Nothing else. Every run replaces that ID list with IDs inside the query window (timestamps at or after lastSeen). When the watch is current, that window is the recent overlap, and the API end bound is today plus two days. IDs older than lastSeen are dropped so the set does not keep the whole history.
• Caps. Extra dockets or filers past the cap are not queried. New filings past maxFilings stay unseen so the next run can return them. A failed query does not move lastSeen.
• DEMO_KEY. On 15 Oct 2026 the demo key's response included x-ratelimit-limit: 10. The header had no reset time, so this README does not state the window. A long list on DEMO_KEY can fail partway; those queries become free error rows.

Input example

{
  "dockets": ["22-270"],
  "filers": [],
  "monitorId": "default",
  "maxFilings": 50
}