Facebook Posts Search Scraper (No Login)

Search public Facebook posts by keyword and get fresh results with text, author, time, links, media and engagement counts. No Facebook account, no cookies and no proxy needed.

Why use it

• No login, no bans. It never signs in, so there are no accounts to get checkpointed and no cookies to refresh.
• Fresh posts. Combines Facebook's recent-posts filter with an age cutoff, then drops anything older than you ask for.
• Batch keywords. Put hundreds of keywords in one run and they're searched in parallel. Far cheaper and faster than one run per keyword.
• Finds what plain search misses. Facebook returns nothing for some exact phrases, such as "plane crash" or "bomb threat". The actor can search reworded variants that still match.
• Drop-in input. Accepts the single-keyword input used by other Facebook search scrapers (keyword, results_limit, since, filter_by_recent_posts), so switching is easy.

Use cases

• Monitoring breaking news, emergencies and local incidents
• Brand and product mentions
• Market and trend research across languages
• Feeding posts into alerting, dashboards or AI pipelines

Input

One keyword:

{
    "keyword": "power outage",
    "results_limit": 25,
    "since": "1d"
}

Many keywords in one run:

{
    "keywords": ["wildfire evacuation", "power outage", "road closed", "inondation"],
    "results_limit": 25,
    "since": "6h",
    "concurrency": 8
}

Field	Default	What it does
keyword / keywords		Search phrase(s). Any language works.
results_limit	25	Maximum posts per keyword
since	1d	Only keep posts newer than 30m, 6h, 1d, 1w. Empty means no cutoff
filter_by_recent_posts	true	Ask Facebook for recent posts
rewrite_blocked_phrases	true	Search reworded variants of phrases Facebook blanks out
max_pages	4	Pages fetched per keyword (about 10 posts per page)
concurrency	8	Keywords searched at the same time
active_until		Optional end date: runs started after it exit without searching

Output

One row per post:

{
    "input": "power outage",
    "search_query": "power outage",
    "post_id": "1234567890123456",
    "create_time": 1791850927000,
    "text": "Crews are working to restore power to about 2,000 customers...",
    "post_url": "https://www.facebook.com/examplepage/posts/pfbid0abc...",
    "author_username": "Example Electric Cooperative",
    "author_user_id": "100000000000000",
    "author_profile_url": "https://www.facebook.com/examplepage",
    "image_list": ["https://scontent.xx.fbcdn.net/..."],
    "video_list": [],
    "type": "photo",
    "like_count": 18,
    "comment_count": 7,
    "share_count": 3
}

create_time is Unix time in milliseconds. search_query differs from input when a blocked phrase was reworded.

Each run also saves a summary to the OUTPUT record in its key-value store: posts per keyword, pages fetched, posts dropped for being too old, and any errors.

Tips

• Batch your keywords. One run with 200 keywords costs a fraction of 200 single-keyword runs.
• Scheduling. For monitoring, schedule a run every 5 to 30 minutes and de-duplicate on post_id. Repeat searches mostly return new posts mixed in with ones you've seen.
• Results order. Facebook ranks results by relevance within the time window, not strictly newest first.
• Memory. 1 GB is the sweet spot. Smaller runs are slower for about the same cost.

Limits

• Only public posts that Facebook shows to logged-out visitors are returned.
• Some phrases are blocked outright and have no working rewrite, for example anything containing "stab" or "suicide".
• Group-only, private and friends-only posts are never included.
• Facebook can change its site at any time. If results drop, the actor will be updated.

Is it legal to scrape Facebook?

This actor only collects publicly visible posts and does not log in. You are responsible for how you use the data, including complying with Facebook's terms, data-protection laws such as GDPR, and any rules that apply to personal data in your jurisdiction. If in doubt, get legal advice.