Migrating from another provider

Switch in three steps

1. 1Get a LurkAPI key. It starts with lk_live_.
2. 2Change the base URL to https://api.lurkapi.com. Paths stay the same.
3. 3Keep sending the key in the x-api-key header.

69 of ScrapeCreators' endpoints are at the same path here, so a client written for it usually works as it is. A call looks like this:

curl "https://api.lurkapi.com/v1/facebook/adLibrary/search/ads?query=running+shoes&country=US&status=active" \-H "x-api-key: YOUR_API_KEY"

Then check your code against the differences below.

What stays the same

• Paths, the x-api-key header, and the success, credits_remaining and credits_charged fields. error is still a string.
• Allowed values work in any case (status=ACTIVE), and so do an r/ prefix on subreddit names and trailing slashes.
• Unknown parameters such as get_transcript are ignored. Blank parameters count as unset, and a repeated parameter uses its first value.

What's different

Requests

• Keys start with lk_live_. Authorization: Bearer <key> works too.
• Only GET. Other methods return 405 method_not_allowed. Some providers also take POST on search ads, company ads and post comments. Send those as GET, with the parameters in the URL.
• Instagram profile, user posts and user reels take username (or url), not handle. For the next page, pass after, not next_max_id or max_id.
• TikTok user_id and region aren't supported on profile, videos, search or shop.
• trim must be true or false (any case). Anything else is 400 invalid_params.
• Search companies country takes ALL or a 2-letter country code, nothing else.
• cache_max_age is ignored. Each endpoint sets its own cache time, shown on its page. A cached answer costs the same as a fresh one.

Responses

• credits_remaining is your real balance after the call.
• Instagram profile and X profile come back in a different shape. Each link shows a real response.
• Facebook Ad Library watermarked_resized_image_url is "" rather than null, and ig_verification is false rather than null.
• Company ads defaults to status=all. Pass status=active for running ads only.
• Subreddit search returns posts only.

Errors and rate limits

• Error bodies are { success: false, error, code, docs, issues? }. Check code, not the error text. 502 and 503 errors don't include the platform's own error details. All error codes.
• Status codes: no credits left is 402 insufficient_credits, too many calls is 429 rate_limited, a missing key is 401 missing_api_key and a revoked key is 401 invalid_api_key.
• Rate limit: 50 requests per 10 seconds per account, with a retry-after header on 429. High-volume accounts get 100. Email support@lurkapi.com to ask.

Charging

A call is charged once it reaches the platform. That includes not_found and other errors the endpoint itself returns, like a bad cursor. These are free: invalid parameters (caught before the call), 401, 402 and 429 rejections, and 5xx failures. A cached answer costs the same as a fresh one. Free tries without an account follow the same rule.

Fields we don't return

Reddit data comes from reddit.com's own pages. We leave out fields those pages don't show rather than guess them. The four Facebook Ad Library endpoints and subreddit search return every key.

• Subreddit posts (posts[]): no downs, subreddit_subscribers, num_crossposts, edited, distinguished, archived, pinned, media, secure_media, media_embed, gilded, all_awardings, author_flair_text or link_flair_css_class, and none of Reddit's moderator (mod_reports, banned_by…) or viewer-state (saved, likes, clicked…) fields.
• Search Reddit (posts[]): only what the search page shows, so also no selftext, selftext_html, url, url_overridden_by_dest, domain, is_self, is_video, preview, upvote_ratio, link_flair_text, stickied, locked or total_awards_received. Pass the post's URL to post comments for the full post.
• Post comments: post lacks the same fields as subreddit posts, plus num_duplicates; comments[] have no downs, edited, distinguished, stickied, locked, controversiality, score_hidden, gilded, all_awardings, total_awards_received, author_flair_text or collapsed_reason, nor moderator or viewer-state fields.
• Subreddit details: no subscribers, advertiser_category or submit_text. Use weekly_active_users for size.

Older `/api/v1/` clients

• /api/v1/… paths still work. https://api.lurkapi.com/api/v1/facebook/adLibrary/search/ads is the same endpoint as https://api.lurkapi.com/v1/facebook/adLibrary/search/ads.
• credits_remaining used to be -1.
• Errors used to be { error }, with different text: "missing required query param: query" is now "query: required". A revoked key used to get 403, and now gets 401.
• The Facebook ad endpoint and the Reddit endpoints used to treat any trim value other than true as false. Now anything but true or false is a 400.

Endpoints and prices

Endpoints cost 1–31 credits per call.

Also included

• Claude and MCP. Every endpoint is also a tool that Claude and other AI apps can call, at https://api.lurkapi.com/mcp. Connect it to Claude.
• Docs for AI agents. llms.txt, openapi.json, and every docs page as markdown (add .md to the URL).
• Pay per call, from $0.59 per 1,000 credits, with no subscription. Pricing.