Government-Filing Delta Monitor — State Lobbying Disclosures

Watches public government filings (state lobbyist disclosures via Socrata SODA open-data APIs) on a user-maintained watchlist and emits one dataset item per newly detected or amended filing. Repeat runs emit only newly detected filings via per-source checkpoints in the Actor key-value store — free when nothing changes, pay only on signal.

Who it is for

Public-affairs firms, compliance officers, government-relations teams and watchdog NGOs that need to track new or amended state lobbying-disclosure filings on a recurring schedule without re-polling open-data portals manually. Different buyer segment from the SEC insider-trading / institutional-holdings monitors — this is state/open-government data, not federal securities filings.

What it monitors

A watchlist of public Socrata SODA endpoints, e.g. the Washington State PDC lobbyist disclosures at https://data.wa.gov/resource/nuwx-ay5h.json. Each source is polled independently; per-source checkpoints remember the last checked high-water date and the filing IDs already emitted, so:

• First run: returns filings within a bounded lookback window (default 30 days).
• Repeat runs: return only newly detected or amended filings (new Socrata row IDs), including amendments that surface under a new ID.
• Nothing changed since last run: zero dataset items — "free when nothing changes."

What makes it different

Existing lobbyist-disclosure scrapers (e.g. titan_coder/wa-lobbyist-reporting-delta at $10.00 per 1,000 new/amended filings, "free when nothing changes"; scrapemint/lobbying-disclosure-scraper) are on-demand fetchers: you run them, get a batch, and run again when you want an update. This Actor keeps a watchlist and remembers what it already reported, so it is a true recurring delta monitor for public affairs / compliance buyers.

Data source

Public Socrata SODA open-government APIs only — no credentials, no proxies, no scraping of rate-limited or anti-bot pages, no commercial inference (no LLMs) inside the product.

Default source (Washington lobbyist disclosures):

• https://data.wa.gov/resource/nuwx-ay5h.json

Each watchlist source needs a public Socrata SODA .json endpoint and is queried with a deterministic GET + $where / $order / $limit filter against a per-source date column (default receipt_date) and ID column (default id).

Pricing (pay-per-event framing, Apify Store)

Per the proven incumbent price point, this Actor frames pricing at:

• $0.01 per detected/amended filing (each dataset item), i.e. $10.00 per 1,000 new or amended filings.
• Actor start / nothing-changed runs: free (no dataset items emitted).

Apify retains a 20% commission on revenue; payouts are monthly and require completed identity verification (KYC). This listing is free until the owner enables monetization and completes payout/KYC — see knowledge/decisions.md (POLICY #6, owner decision 2026-10-30). Pricing shown here is a README framing only; no monetization is flipped in this release candidate.

Input

See actor/.actor/input_schema.json. Required: sources (non-empty array of Socrata SODA endpoints).

• sources[].name: unique handle used as the checkpoint key (e.g. "wa-lobbyist").
• sources[].url: public Socrata SODA JSON endpoint (e.g. https://data.wa.gov/resource/nuwx-ay5h.json).
• sources[].lookbackDays (1-90, default 30): first-run lookback window.
• sources[].dateColumn (default "receipt_date"): Socrata column used to order/filter filings by recency.
• sources[].idColumn (default "id"): Socrata column that uniquely identifies each filing (dedupe key).
• maxItems (1-200, default 50): cap on items emitted per source per run.

Example (the default input — returns WA lobbyist filings from the last 30 days):

{
  "sources": [
    {
      "name": "wa-lobbyist",
      "url": "https://data.wa.gov/resource/nuwx-ay5h.json",
      "lookbackDays": 30
    }
  ],
  "maxItems": 50
}

Output

One dataset item per newly detected or amended filing. Every field is present; fields that could not be parsed are null — the Actor never guesses.

{
  "filingId": "2073682",
  "filingDate": "2026-10-31",
  "filerName": "ROB MAKIN CONSULTING*",
  "filerType": "Firm",
  "clientName": null,
  "subject": null,
  "amount": null,
  "documentUrl": "https://my.pdc.wa.gov/lobbyist/public/-/#/public/l2-report/2073682",
  "source": "wa-lobbyist",
  "newSinceLastRun": true,
  "amendedByReport": null,
  "year": "2026",
  "reportFrom": "2026-10-01",
  "reportThrough": "2026-10-31"
}

Limits

• Max 200 items per source per run (maxItems).
• First-run lookback capped at 90 days; repeat runs are bounded by the per-source checkpoint date.
• Dedupe is by Socrata row ID within a capped seen-IDs window (5,000 IDs per source). Resetting the Actor key-value store re-emits filings within the lookback window.
• One erroring source never aborts the whole run.

Verification

• pytest actor/tests — deterministic unit tests (no network): input normalization, row-to-item mapping, first-run vs. repeat-run checkpoint behavior, dedupe, and max-items cap.
• ruff check actor/ — must be clean (0 errors).
• Default-input smoke run against the live Socrata endpoint returns more than 0 items in under 5 minutes (publish precondition).