Discourse Forum Monitor — New Topics & Replies

Know the moment someone mentions your product in a community — without reading the forum.

Almost every developer tool and SaaS company runs its community on Discourse: OpenAI, Docker, Home Assistant, Cloudflare, Ghost, Rclone, and thousands more. This actor watches the ones you pick and reports only what changed since your last run.

Signal	What it means
🆕 New topic	Someone started a thread. If you're watching a keyword, they just mentioned you
💬 New replies	The thread is alive — and how many posts were added
✅ Answer accepted	The question got resolved
📈 Trending	Views are spiking on an existing thread
🔒 Closed	Moderators locked it
❌ Removed	It's gone

If nothing changed, you get nothing and you pay nothing per result.

Two ways to use it

Brand monitoring — give it keywords and it watches every mention across every forum you list:

{
  "forumUrls": ["community.openai.com", "meta.discourse.org"],
  "keywords": ["your-product", "competitor-name"],
  "monitorKey": "brand-watch"
}

Community pulse — leave keywords empty and it watches the forum's recent activity:

{
  "forumUrls": ["community.openai.com"],
  "maxPages": 3,
  "trackSignals": ["new_topic", "trending"]
}

Why not just scrape the forum

There are Discourse scrapers already. They hand you the same 500 topics every run and leave you to work out what's different. For a daily schedule that's 500 rows you already had, every day.

Run over 2 forums → 100 topics watched → 3 rows returned:
  [new_topic  ] New Realtime API voices and cache pricing
  [new_replies] Realtime API updates — WebRTC        +3 replies
  [answered   ] Anyone know the pricing of full o1 image

You pay for 3 rows, not 100. The comparison happens against a snapshot stored in your own Apify account.

Built for schedules

Set it to run hourly or daily and let it report.

• DevRel & community managers — never miss a question about your product
• Marketing — track how a competitor is discussed, in their own community
• Support teams — catch the unanswered thread before it becomes a complaint
• Researchers — build a real time series of community activity

First run is free

The first run records the current state as your baseline and charges nothing per result. It returns a 25-row sample so you can see the data shape. From the second run onward you see and pay for changes only.

Running several watchlists

monitorKey is the memory. Different keys, independent histories:

{ "monitorKey": "our-brand",  "keywords": ["acme"] }
{ "monitorKey": "competitor", "keywords": ["globex"] }

Keep the key stable. Change it and you start a fresh baseline.

Just need the topics right now?

Set oneShot: true and it returns the current topics instead of what changed — no history, no waiting for a second run.

{ "oneShot": true, "forumUrls": ["community.openai.com"], "keywords": ["pricing"] }

Use it for a one-off lookup, to see real data before committing to a schedule, or when calling from an AI agent. It never touches your saved monitoring history, so it is safe to run against a monitorKey you already use.

Honest limits

• Keyword search does not return view counts. Discourse's search.json omits them, so the trending signal and the minViews filter only work when watching recent topics (no keywords).
• Keyword search returns up to 50 topics per keyword per forum.
• Changing maxPages or your keyword list between runs changes what is being watched, which shows up as a burst of new and removed topics. Keep them stable.
• On removed topics only the stored summary survives — the forum no longer serves them.
• Private categories and login-walled forums are invisible. Only public content is read.

Where the data comes from

The same public JSON endpoints the forum's own front end uses: /latest.json for recent topics and /search.json for keyword search. No login, no API key, no scraping of rendered HTML. No personal data beyond the public author username of the last post.