Substack New-Post Monitor

Watches your favorite Substack newsletters and reports only genuinely new posts. Schedule it daily or hourly: runs where nothing new was published push zero rows and cost nothing beyond the Actor start — you pay per new post detected, never per scan.

How it works

1. For each publication you configure, the Actor fetches the latest public posts from Substack's public API (/api/v1/archive) — no auth, no key, no proxy.
2. It compares them against a saved snapshot in a named key-value store (substack-monitor-state), namespaced per stateKey, so separate monitors never share or clobber baselines.
3. Posts it has never seen are pushed to the dataset and billed once each (new-substack-post event). Everything else — including scheduled re-runs with zero new posts — pushes nothing and bills nothing.
4. Only public post metadata is read (title, subtitle, public excerpt, published date, author, URL, audience). For paywalled posts the Actor sets paywalled: true and reports the public metadata only — it never attempts to read paid content.

Input

{
  "publications": [
    { "url": "https://platformer.substack.com", "maxPosts": 10 },
    { "url": "https://astralcodexten.substack.com", "maxPosts": 20 }
  ],
  "stateKey": "default",
  "reportInitialAsNew": true
}

• publications — one entry per newsletter. Use the publication's substack.com URL, e.g. https://platformer.substack.com. Custom-domain publications serve their public API on the underlying substack.com URL, so use that instead.
• stateKey — namespace for the snapshot. Change it to start a fresh baseline. The prefilled default key rotates daily so the free trial always shows recent posts; use your own key for persistent monitoring.
• reportInitialAsNew — first run reports everything as new (builds the baseline while reporting); turn off to silently establish the baseline.

Output (dataset)

One row per genuinely new post:

field	meaning
publicationName, publicationUrl	which newsletter
postId	stable Substack post id (dedup key)
title, subtitle, excerpt	public metadata
url	canonical post URL
publishedDate	ISO timestamp
author, authorHandle	byline
postType	e.g. newsletter, podcast
paywalled	true when audience is not everyone (public metadata only)
detectedAt	when this Actor first saw it

A run receipt (substack-monitor-state → state-v1-<stateKey>-last-receipt) records per-publication status, fetch counts, and errors for auditing.

Pricing (pay-per-event)

• Actor start: $0.001 (tiered: Bronze $0.0009, Silver $0.0008, Gold/Platinum/Diamond $0.0007)
• New post detected: $0.0025 (tiered: Bronze $0.0023, Silver $0.0021, Gold/Platinum/Diamond $0.0020)

Storefront shows "from $2.00 / 1,000 new posts". Compare with one-shot scrapers: a daily scheduled one-shot re-run bills per scanned post every day; this monitor bills only when there is genuinely something new — a quiet week costs ~$0.007.

A daily monitor that finds 3 new posts in a week costs roughly 7 × $0.001 + 3 × $0.0025 = $0.0145. Runs that find nothing cost just the start event. (All prices are the FREE-tier event price; Apify's tiered discounts apply at higher plans.)

Limits & honesty

• Public API only. If Substack removes or gates an endpoint, affected publications report status: "error" in the run receipt instead of failing the run.
• Invalid URLs and nonexistent publications are reported per-publication with a clear message; the rest of the run continues.