60 Claude Prompts That Use Live Web Data, Sorted by Role

Contains affiliate links. Disclosure.

Quick Answer

Each prompt below tells Claude which Apify Actor to run, what to pull, and what to hand back. Paste it into Claude with the Apify connector switched on. Every prompt makes Claude check the Actor's inputs and price first and wait for your OK before it spends anything. Each one links to a full guide where we ran the workflow for real on 2026-10-21.

Generic prompt lists ask Claude to "research competitors" and get an answer from training data. These prompts name a real Actor, real input fields and the traps our test runs hit, so Claude works from today's web, not last year's.

Before you paste: connect Apify once

Every prompt assumes Claude can reach the Apify MCP server. Pick the guide for your client and do it once:

Client	Setup guide
Claude Cowork (most prompts here)	Connect Apify to Cowork: Customize > Connectors, add https://mcp.apify.com
Claude Desktop chat	Apify MCP in Claude Desktop
Claude Code	Claude Code web research setup
New to Claude Code	Claude Code for non-coders

Three rules are written into every prompt, so you can paste any of them on its own:

1. Check before spending. Claude calls fetch-actor-details to read the Actor's current inputs and price, tells you the maximum cost, and waits for your OK.
2. Small test first. The first run is capped at a handful of results.
3. No guessing. Missing data is marked "unknown", and every claim cites the row or page it came from.

Prompts marked weekly or daily work as Cowork scheduled tasks. Paste them as the task prompt once you're happy with a manual run.

What the data costs

Most prompts cost cents per run. These are the free-plan prices our test runs were checked against on 2026-10-21; paid Apify plans pay less per result.

Actor	Used for	Price per result (free plan)
LinkedIn Jobs Scraper	Job ads	$0.002 subject to change · verified 2026-10-21
Indeed Jobs Scraper	Job ads	$0.0001 subject to change · verified 2026-10-21
Google Maps Reviews Scraper	Google reviews	$0.0006 subject to change · verified 2026-10-21
Amazon Reviews Scraper	Amazon reviews	$0.006 subject to change · verified 2026-10-21
Airbnb Scraper API	Airbnb listings	$0.003 subject to change · verified 2026-10-21
YouTube Scraper	Video stats	$0.004 subject to change · verified 2026-10-21
YouTube Transcript Scraper	Transcripts	$0.001 subject to change · verified 2026-10-21
Reddit Scraper Lite	Reddit posts and comments	$0.004 subject to change · verified 2026-10-21

Some Actors also charge a small start fee per run. The AI-search, SERP and B2B-signal Actors are priced in their own guides. Prices change, which is why every prompt has Claude read the live price before running.

Recruiters

The full workflow, including the legal limits on scraping personal data, is in Claude for recruiting: build a candidate shortlist. It uses job ads and companies' own team pages only, never LinkedIn profiles.

1. Companies hiring for a role like minePaste into Claude Cowork

Find companies that posted job ads for "[role title]" in [location] in the last 30 days.
Use the Apify Actors curious_coder/linkedin-jobs-scraper (job ads only, never profiles) and valig/indeed-jobs-scraper, 25 results each.
Before any paid run, check each Actor's inputs and price with fetch-actor-details, tell me the maximum cost, and wait for my OK.
Return one table, one row per company: company, open roles (merge the same title posted in several cities), locations, newest posting date, ad links.

Full workflow: build a candidate shortlist in Claude

2. Public team pages for target companiesPaste into Claude Cowork

For these company websites: [domain 1], [domain 2], [domain 3]
use apify/website-content-crawler to read only their team, about or leadership pages (stay on each domain, max 5 pages per site).
Before running, check the Actor's inputs and price with fetch-actor-details, tell me the maximum cost, and wait for my OK.
Return: company, person's public name, title as written on the page, and the page URL as the source.
Do not visit LinkedIn or GitHub profiles. Do not guess anything that is not on the page; write "unknown".

Full workflow: build a candidate shortlist in Claude

3. Score a shortlist against the role specPaste into Claude

Here is my role spec: [paste must-haves and nice-to-haves].
Here is my candidate table: [paste or attach].
Score each person against the written must-haves only. For every score, quote the source text that supports it.
If there is no evidence for a must-have, mark it "unknown", not zero.
Never infer age, gender, ethnicity or any other protected trait from names, photos or anything else.
End with a ranked list and a note of what I should verify by hand. Do not run any Apify Actors.

Full workflow: build a candidate shortlist in Claude

4. Outreach drafts for approved candidates onlyPaste into Claude

Draft a short first message to each person I marked "approved" in this table: [paste].
Each draft must say where I found their name (the exact public page), why the role fits, and how to opt out of further contact.
Keep it under 120 words, no flattery, no claims about them that are not in the source.
Do not draft anything for people I have not approved.

Full workflow: build a candidate shortlist in Claude

5. Check the pay band before you open the rolePaste into Claude Cowork

Benchmark pay for "[job title]" in [city, country].
Use valig/indeed-jobs-scraper (up to 100 ads, posted in the last 14 days). Before running, check its inputs and price with fetch-actor-details, tell me the maximum cost, and wait for my OK.
Keep only ads whose title really matches; list the ones you excluded and why.
Use the posted salary range midpoints, annualise hourly pay at x2080, drop ranges stated for a different location.
Report P25, P50 and P75 with the sample size n, and the five ads closest to the median as evidence.

Full workflow: salary benchmarking with Claude

6. Who is competing with me for this talentPaste into Claude Cowork

Count how many job ads each company posted for the "[role family, e.g. data engineering]" family in [location] in the last 30 days.
Use curious_coder/linkedin-jobs-scraper (job ads only), 50 results. Check inputs and price with fetch-actor-details first and wait for my OK.
Merge duplicate ads (same title in several cities counts once).
Return the top 10 companies by open roles, with seniority mix where the ad states it, and one line on what that means for my hiring.

Full workflow: build a candidate shortlist in Claude

Job seekers

The scheduled version, with the CV pasted into the task so it runs while your laptop is closed, is in Claude job search: a daily digest ranked against your CV.

7. Today's new jobs, ranked against my CVPaste into Claude Cowork

My CV: [paste as text].
My filters: title "[title]", location "[city or remote]", must-haves [list], deal-breakers [list].
Find postings from the last 24 hours with curious_coder/linkedin-jobs-scraper and valig/indeed-jobs-scraper, 25 each.
Before running, check inputs and price with fetch-actor-details, tell me the maximum cost, and wait for my OK.
Drop duplicates and keep at most 2 postings per company. Score each against my CV and filters.
Return the top 5 with the link, a fit score, and 3 tailored CV bullets each. Never invent experience I do not have.

Full workflow: daily job search digest with Claude

8. Is this offer fair?Paste into Claude Cowork

I have an offer for "[title]" in [city] at [amount]. Check it against live job ads.
Use valig/indeed-jobs-scraper (up to 100 ads, last 14 days). Check inputs and price with fetch-actor-details first and wait for my OK.
Keep only true title matches, use range midpoints, and report P25/P50/P75 with n.
Tell me where my offer sits and which matching ads pay more, with links.

Full workflow: salary benchmarking with Claude

9. Tailor my CV to one postingPaste into Claude Cowork

My CV: [paste]. The job: [posting URL].
Fetch the posting with apify/rag-web-browser (1 page). Check its inputs and price with fetch-actor-details first and wait for my OK.
List the posting's stated requirements, then show which ones my CV already proves (quote the line) and which it does not.
Rewrite my top 5 bullets to match, using only facts already in my CV.

Full workflow: daily job search digest with Claude

10. Which skills keep coming up?Paste into Claude Cowork

Pull 50 recent job ads for "[title]" in [location] with curious_coder/linkedin-jobs-scraper.
Check inputs and price with fetch-actor-details first and wait for my OK.
Count how often each skill, tool and certification is mentioned across the ad descriptions.
Compare the top 15 with my CV: [paste]. Return a table of skill, ads mentioning it, and whether my CV shows it.

Full workflow: daily job search digest with Claude

11. Brief me on a company before the interviewPaste into Claude Cowork

I am interviewing at [company] ([domain]) for [role] on [date].
Use apify/rag-web-browser with 2 or 3 searches (recent company news, the product, the team I would join). Check inputs and price with fetch-actor-details first and wait for my OK.
Write a one-page brief: what they sell, recent news from the last 90 days, likely priorities for this role, and 5 sharp questions to ask.
Cite a source URL for every fact. Say "not found" rather than guess.

Full workflow: daily job search digest with Claude

12. Tune my digest after a weekPaste into Claude Cowork

Here are this week's job digest reports: [paste].
For each posting I applied to, I marked "applied"; the rest I skipped.
Work out what my skips have in common (title drift, seniority, location, company type) and rewrite my filters and scoring rules so next week's top 5 is closer to what I actually apply to.
Do not run any Apify Actors. Show the old and new rules side by side.

Full workflow: daily job search digest with Claude

SEO

These prompts read the live Google results and AI answers. The brief workflow is written for Claude Code, where the brief lands in your repo as a file.

13. Content brief from the live top 10Paste into Claude Code

Build a content brief for the keyword "[keyword]" (country [US], language [en]).
Use apify/google-search-scraper with its full website content option so one run returns the results and the text of each ranking page.
Check inputs and price with fetch-actor-details first, tell me the maximum cost, and wait for my OK. Cap the run at $0.25.
From the data: list each ranking page's H2s and word count, the People Also Ask questions and related searches, the subtopics most pages cover, and gaps with evidence (how many pages mention each term).
Flag any page whose fetch looks thin. Ask before saving the brief to briefs/[keyword].md.

Full workflow: Claude Code SEO content briefs

14. People Also Ask harvest for a topicPaste into Claude Code

For these 5 keywords: [list], run apify/google-search-scraper (results only, no page content) for [country].
Check inputs and price with fetch-actor-details first and wait for my OK.
Collect every People Also Ask question and related search, remove near-duplicates, and cluster them by intent.
Return the clusters with the keywords each question came from, and mark which questions my page [URL] already answers.

Full workflow: Claude Code SEO content briefs

15. Refresh check for a page that is slippingPaste into Claude Code

My page [URL] targets "[keyword]". Run apify/google-search-scraper with full website content for that keyword.
Check inputs and price with fetch-actor-details first and wait for my OK.
Compare my page's headings and topics with the current top 10. List subtopics that at least 4 ranking pages cover and mine does not, with the count as evidence.
Do not suggest changes that are not backed by the data.

Full workflow: Claude Code SEO content briefs

16. Title and description patterns that rankPaste into Claude Code

Run apify/google-search-scraper (results only) for "[keyword]" in [country].
Check inputs and price with fetch-actor-details first and wait for my OK.
List the top 10 titles and descriptions. Describe the patterns (length, year, numbers, brackets, angle).
Then propose 3 titles under 60 characters and 2 descriptions under 155 characters for my page about [topic], each different from what already ranks.

Full workflow: Claude Code SEO content briefs

17. Does Google's AI Overview cite me?Paste into Claude Cowork

For these 10 keywords: [list], run apify/google-ai-overviews-scraper for country [US].
Check inputs and price with fetch-actor-details first, tell me the maximum cost, and wait for my OK.
For each keyword: does an AI Overview appear, which domains it cites, and whether [my domain] is cited.
Return a table and a one-line summary of which competitors are cited most.

Full workflow: track AI-search mentions with Claude

18. Is ChatGPT recommending us?Paste into Claude Cowork

Run these 10 buyer questions: [list] through apify/chatgpt-search-scraper and apify/perplexity-search-scraper (countryCode [US]).
Check inputs and price with fetch-actor-details first, tell me the maximum cost, and wait for my OK. Test with 2 questions first.
For each answer, record which brands are named in the text and which domains appear in the sources. A brand cited in sources but not named in the text counts separately.
Return share of voice for [my brand] vs [competitors] per engine.

Full workflow: track AI-search mentions with Claude

Sales

Two guides cover sales: buying signals with Claude for named accounts, and the AI lead generation playbook for building a local list from Google Maps.

19. Local lead list from Google MapsPaste into Claude Cowork

Build a lead list of "[business type]" in [city, state], 20 places.
Use compass/crawler-google-places. Check inputs and price with fetch-actor-details first, tell me the maximum cost, and wait for my OK. Every extra filter is billed per place, so ask before adding one.
Return name, address, phone, website, rating and review count.
Strip utm_ and other tracking parameters from websites and dedupe on the bare domain. Save as CSV after I confirm.

Full workflow: AI lead generation playbook

20. Which of my accounts are hiring?Paste into Claude Cowork

For these accounts: [company names], find job ads posted in the last 30 days with curious_coder/linkedin-jobs-scraper.
Check inputs and price with fetch-actor-details first and wait for my OK.
Merge the same title posted in several cities into one role. Flag roles that suggest a buying need for [what I sell], and quote the line in the ad that shows it.

Full workflow: buying signals with Claude

21. Funding and news triggersPaste into Claude Cowork

Search recent news for each of these accounts: [company names] with data_xplorer/google-news-scraper-fast, last 30 days, 10 results each.
Check inputs and price with fetch-actor-details first and wait for my OK.
Keep only items that are really about that company: drop company-profile pages, other companies with a similar name, and matches on the name as an ordinary word.
List funding rounds, leadership changes, launches and expansions with the source link. Say "none found" when that is the answer.

Full workflow: buying signals with Claude

22. What tech is this account running?Paste into Claude Cowork

Look up the website technology of these domains: [list] with automation-lab/wappalyzer-technology-lookup.
Check inputs and price with fetch-actor-details first and wait for my OK.
Ignore any detection with confidence under 90.
Flag accounts using [competitor tool] or missing [category I sell into], and list the evidence for each.

Full workflow: buying signals with Claude

23. Score signals and log them in HubSpotPaste into Claude Cowork

Here are this month's hiring, news and tech-stack findings for my accounts: [paste].
Score each account 0 to 10 for buying intent using these rules: [your rules]. Show the evidence behind every point.
Using the HubSpot connector, add a note to the existing company record for each account scoring 6 or more. Do not create new companies or contacts.
Show me every note before you write it and wait for my OK.

Full workflow: buying signals with Claude

24. One-page brief before a sales callPaste into Claude Cowork

I have a call with [company] ([domain]) about [what I sell].
Use apify/rag-web-browser with 2 or 3 searches (their product, recent news, open roles). Check inputs and price with fetch-actor-details first and wait for my OK.
Write a one-page brief: what they do, recent changes, likely pains related to [my product], and 4 discovery questions.
Cite a source for every fact and say "not found" rather than guess.

Full workflow: buying signals with Claude

Product managers

For software products. Physical-product reviews are under e-commerce below. The full coding method is in voice of customer analysis with Claude.

25. Competitor complaints on G2Paste into Claude Cowork

Pull recent G2 reviews for [competitor product] with zen-studio/g2-reviews-scraper (limit at least 10, start with 20).
Check inputs and price with fetch-actor-details first and wait for my OK.
Code each "dislike" into themes. Report how many reviews mention each theme, and keep the review source and incentivized flag so I can check for review-campaign bias.
Return the themes ranked by frequency with 2 short verbatim quotes each, no reviewer names.

Full workflow: voice of customer analysis with Claude

26. 1 to 3 star App Store reviewsPaste into Claude Cowork

Pull the latest 1 to 3 star App Store reviews for [app name or URL] with thewolves/appstore-reviews-scraper, 50 reviews, country [us].
Check inputs and price with fetch-actor-details first and wait for my OK.
Group them into themes, separate bugs from missing features, and note which app version each complaint mentions where stated.

Full workflow: voice of customer analysis with Claude

27. Why people switch, from RedditPaste into Claude Cowork

Search Reddit for "[competitor] alternative" and "switched from [competitor]" with trudax/reddit-scraper-lite, 30 items, at most 5 items per thread.
Check inputs and price with fetch-actor-details first and wait for my OK.
Keep only comments where someone explains why they left or chose a tool. Return the reasons ranked by frequency with short quotes and thread links, no usernames.

Full workflow: voice of customer analysis with Claude

28. Comparison videos, read for youPaste into Claude Cowork

Find 5 YouTube videos titled like "[my product] vs [competitor]" or "[competitor] review" with streamers/youtube-scraper.
Then pull their transcripts with codepoetry/youtube-transcript-ai-scraper.
Check both Actors' inputs and prices with fetch-actor-details first, tell me the total maximum cost, and wait for my OK.
Extract every stated reason for preferring one product, with the video and timestamp.

Full workflow: voice of customer analysis with Claude

29. Turn coded feedback into a gap tablePaste into Claude

Here is my coded feedback from G2, the App Store, Reddit and YouTube: [paste or attach].
Build a gap table: theme, sources mentioning it, count, severity (blocker, painful, annoyance) with the rule you used, and whether my product already solves it ([paste my feature list]).
Rank by count times severity. Add 2 verbatim quotes per row for the PRD. Do not run any Apify Actors.

Full workflow: voice of customer analysis with Claude

30. Monthly refresh: only what is newPaste into Claude Cowork

Last month's coded feedback is in [file]. Pull only reviews and threads newer than [date] from the same sources and Actors as last time.
Check inputs and price with fetch-actor-details first and wait for my OK.
Code the new items with the same theme list, add new themes only if 3 or more items need one, and show what moved up or down since last month.

Full workflow: voice of customer analysis with Claude

Marketers

Brand mentions are split by surface: social platforms here, AI answers under SEO. The Reddit workflow is in Reddit brand monitoring in Claude Cowork.

31. Reddit mentions, labelled before countedPaste into Claude Cowork

Search Reddit for "[brand]" and its spellings [list] from the past week with trudax/reddit-scraper-lite, 25 items. Search r/[my subreddit] separately.
Check inputs and price with fetch-actor-details first and wait for my OK.
Label every item first: real conversation about us, hiring or announcement, link to something built with us, unrelated meaning of the word, or spam. Give sentiment only to real conversations.
Count mentions only after labelling. No usernames in the output.

Full workflow: Reddit brand monitoring in Claude Cowork

32. Threads worth answeringPaste into Claude Cowork

From this week's labelled Reddit mentions: [paste], flag the threads where a reply from us would genuinely help (an unanswered question, a wrong fact, a comparison we are part of).
Give a one-line reason for each. For the top 3, draft a reply I can edit that says I work for [brand], answers the question first, and does not pitch.
Never post anything yourself.

Full workflow: Reddit brand monitoring in Claude Cowork

33. TikTok and YouTube mentions this monthPaste into Claude Cowork

Search TikTok for "[brand]" with clockworks/tiktok-scraper and YouTube with streamers/youtube-scraper, 20 results each, last 30 days.
Check both Actors' inputs and prices with fetch-actor-details first and wait for my OK.
Keep only videos that are actually about us. Return creator handle, views, date, link, and one line on what they said. Flag anything negative with more than [N] views.

Full workflow: Reddit brand monitoring in Claude Cowork

34. Weekly AI share of voicePaste into Claude Cowork

Run my 20 buyer prompts: [list] through apify/chatgpt-search-scraper, apify/perplexity-search-scraper and apify/google-ai-overviews-scraper.
Check inputs and prices with fetch-actor-details first, tell me the total maximum cost, and wait for my OK.
Return a table per engine: prompts where [brand] is named, where competitors are named, and where we are cited as a source but not named. Compare with last week's file if I attach it.

Full workflow: track AI-search mentions with Claude

35. Competitor messaging snapshotPaste into Claude Cowork

Read the homepage and pricing page of [competitor 1], [competitor 2] and [competitor 3] with apify/website-content-crawler (max 2 pages per site).
Check inputs and price with fetch-actor-details first and wait for my OK.
Compare headline promise, target customer, proof points, pricing model and calls to action in one table, quoting the page. End with 3 positioning gaps we could own.

Full workflow: MCP servers for small business: the 3 gaps

36. Turn a YouTube channel into a cited knowledge basePaste into Claude Cowork

Pull transcripts for the latest [20] videos of [channel URL] with codepoetry/youtube-transcript-ai-scraper (free captions first; ask me before using the paid AI fallback).
Then split them with codepoetry/markdown-rag-chunker.
Check both Actors' inputs and prices with fetch-actor-details first, run a 3-video test, show me the cost, and wait for my OK before the full run.
Save the chunks to [folder] with each video's title, URL and timestamps so every answer can cite the video.

Full workflow: chat with any YouTube channel in Claude

E-commerce

Claude's Shopify connector reads your own store. Competitors' listings and reviews still need scraping, which is what these prompts do. The method is in Claude for Amazon and Shopify sellers.

37. Competitors' 1 to 3 star Amazon reviewsPaste into Claude Cowork

Pull critical reviews for these competitor ASINs: [list] with junglee/amazon-reviews-scraper. Use the "critical" star filter, sort by recent, 20 reviews per product, and set a 180-second timeout.
Check inputs and price with fetch-actor-details first, tell me the maximum cost, and wait for my OK.
Drop userId and profile link columns. Code the complaints into themes (product, packaging, instructions, listing accuracy) and rank by frequency with short quotes.

Full workflow: Claude for Amazon and Shopify sellers

38. Competitor Shopify store reviewsPaste into Claude Cowork

Pull reviews rated 3 stars or lower from [competitor store URL] with stanvanrooy6/shopify-review-scraper.
Check inputs and price with fetch-actor-details first and wait for my OK. If it returns nothing, tell me the store's review app may not be supported rather than retrying.
Drop reviews with no real text. Code the rest into themes with counts and short quotes, no reviewer names.

Full workflow: Claude for Amazon and Shopify sellers

39. Fix list for my own listingPaste into Claude Cowork

Here are my competitors' coded complaints: [paste]. Here is my listing: [paste title, bullets and description, or read it with the Shopify connector].
For each top complaint theme, say whether my listing already answers it, and if not, write the exact bullet or sentence to add.
Only promise what my product actually does: ask me when you are not sure. Do not run any Apify Actors.

Full workflow: Claude for Amazon and Shopify sellers

40. Packaging and gifting complaintsPaste into Claude Cowork

From these competitor reviews: [paste or attach], pull out every complaint about packaging, damage in transit, missing parts or gifting (bought as a gift, recipient's experience).
Group them, count them, and suggest 3 packaging or insert changes that would prevent the top issues, each tied to the quotes behind it.

Full workflow: Claude for Amazon and Shopify sellers

41. Competitor product page snapshotPaste into Claude Cowork

Read these competitor product pages: [URLs] with apify/e-commerce-scraping-tool.
Check inputs and price with fetch-actor-details first and wait for my OK.
Return title, price, rating, review count and key claims for each in one table, then list claims they make that my listing [paste] does not.

Full workflow: MCP servers for small business: the 3 gaps

42. Monthly review refreshPaste into Claude Cowork

Last month's coded reviews are in [file]. Pull this month's critical reviews for the same ASINs with junglee/amazon-reviews-scraper, sorted by recent.
Do not use the date filter (it adds a charge): fetch recent reviews and drop anything already in last month's file.
Check inputs and price with fetch-actor-details first and wait for my OK.
Show new themes and themes that grew, with quotes.

Full workflow: Claude for Amazon and Shopify sellers

Local businesses

Claude drafts, you post. Replies go out from your Google Business Profile, and suspected fake reviews are a human decision under Google's policy. The full workflow is in respond to Google reviews with Claude.

43. Draft replies to last week's reviewsPaste into Claude Cowork

For my locations: [Google Maps place URLs or place IDs], pull reviews from the last 8 days with compass/google-maps-reviews-scraper.
Check inputs and price with fetch-actor-details first, tell me the maximum cost, and wait for my OK.
Skip reviews that already have an owner reply. For 3 stars or lower, alert me first. Draft replies in this voice: [paste 2 replies I liked]. Max 80 words, no reviewer names, sign off as [name].
Do not draft anything for reviews that mention [legal, injury, refund over $X]: list those for me instead. Never post replies yourself.

Full workflow: respond to Google reviews with Claude

44. Make sure we are scraping the right branchPaste into Claude Cowork

Here are my locations as I know them: [names and addresses]. For each, run compass/google-maps-reviews-scraper with max 3 reviews on the place URL I give you: [URLs].
Check inputs and price with fetch-actor-details first and wait for my OK.
Confirm the name and address returned match my branch. If a URL lands on a different place, say so and ask me for the exact place URL. Then list the place IDs to reuse.

Full workflow: respond to Google reviews with Claude

45. Our rating vs nearby competitorsPaste into Claude Cowork

Find "[my category]" businesses within [area] with compass/crawler-google-places, 15 places, including mine: [my business name].
Check inputs and price with fetch-actor-details first and wait for my OK.
Rank by rating and review count, show where we sit, and list what the top 3 do differently that is visible in their listing.

Full workflow: respond to Google reviews with Claude

46. Check a suspicious burst of reviewsPaste into Claude

Here are this month's reviews for [location]: [paste]. My normal volume is about [N] a week.
Look for patterns that could suggest a review burst: unusual volume in a short window, similar wording, very new accounts, no text.
Present evidence only, with dates and counts. Do not conclude that any review is fake: I decide, and I report through Google's own process if needed.

Full workflow: respond to Google reviews with Claude

47. Find broken links on my websitePaste into Claude Code

Check [my site URL] for broken links with khadinakbar/broken-link-checker, capped at 200 links.
Check inputs and price with fetch-actor-details first, tell me the maximum cost, and wait for my OK.
Return each broken link, the page it is on, and the status code, grouped by page. Suggest the likely correct URL where it is obvious.

Full workflow: Claude Code for non-coders

48. Which connectors should my business use?Paste into Claude Desktop

I run a [type of business] with [N] staff. We use [tools: email, calendar, accounting, CRM, store].
Recommend 3 to 5 Claude connectors from the official directory for us, one sentence each on what it would do for us.
Then tell me which jobs no connector covers (for example competitor prices or Google reviews) and which Apify Actor would fill each gap. Do not connect or run anything yet.

Full workflow: MCP servers and connectors for small business

Real estate

Short-term rental numbers come from calendar-implied occupancy, explained once in the Airbnb comp guide. The investor workflow is in STR vs LTR yields with Claude. None of this is an appraisal.

49. Weekly Airbnb comp memoPaste into Claude Cowork

My listing: [Airbnb URL], [N] bedrooms, in [city]. Find comparable listings with codepoetry/airbnb-scraper: search the city, then keep only comps within [X] km of my listing by latitude and longitude (the Actor has no radius input).
Run one search per weekend I care about: [dates]. Check inputs and price with fetch-actor-details first, run a 5-listing test, and wait for my OK.
Report each comp's nightly rate and occupancy, where I am priced below the market, and amenities guests praise in comps' reviews.

Full workflow: price your Airbnb against nearby comps

50. How booked are my competitors?Paste into Claude Cowork

For these comparable Airbnb listings: [URLs], pull the availability calendar for the next 3 months with codepoetry/airbnb-scraper.
Check inputs and price with fetch-actor-details first and wait for my OK.
Report blocked days as a share of days in the window for each listing. Remind me that blocked days include owner blocks, not only bookings.

Full workflow: price your Airbnb against nearby comps

51. Homes for sale in a ZIP codePaste into Claude Cowork

List homes for sale in ZIP [code] under $[budget] with maxcopell/zillow-zip-search, 20 results.
Check inputs and price with fetch-actor-details first and wait for my OK.
Return address, price, beds, baths, square feet and price per square foot, sorted by price per square foot.

Full workflow: STR vs LTR yields with Claude

52. Long-term rent compsPaste into Claude Cowork

List rentals in ZIP [code] with [N] bedrooms using maxcopell/zillow-zip-search in rent mode, 20 results.
Check inputs and price with fetch-actor-details first and wait for my OK.
Rents shown as "from" prices like "$1,299+" are floors, not typical rents: mark them. Give me the median rent with and without the floor prices.

Full workflow: STR vs LTR yields with Claude

53. Short-term vs long-term yield per propertyPaste into Claude

Here are the properties I am considering: [paste], short-term rental comps: [paste], and long-term rent comps: [paste].
My assumptions: management fee [%], cleaning per stay $[X], vacancy for long-term [%], annual costs $[X]. Label each one as my assumption.
For each property, show gross and net yield for short-term and long-term letting, with the formula. Flag where the answer flips depending on the assumptions. This is not an appraisal.

Full workflow: STR vs LTR yields with Claude

54. Flag the outliersPaste into Claude

From this ZIP-level listing table: [paste], compute the median price per square foot and flag any listing more than 25% above or below it.
For each outlier, list what in the data could explain it (size, condition notes, lot, property type) and what I should check in person. Do not run any Apify Actors.

Full workflow: STR vs LTR yields with Claude

Creators

An outlier here is a video's views divided by the median views of its own channel's videos in the window, excluding Shorts and videos less than 7 days old. The method is in find outlier videos in your niche.

55. Outlier videos in my nichePaste into Claude Cowork

For these YouTube channels: [URLs], pull videos from the last 90 days with streamers/youtube-scraper (no Shorts).
Check inputs and price with fetch-actor-details first, test with 2 channels, and wait for my OK. Tell me if any handle does not exist.
Score each video: views divided by its own channel's median views, using only videos at least 7 days old, and only channels with at least 5 such videos.
Return every video at 3x or more, sorted by score, with title, views, score and link.

Full workflow: find outlier videos in your niche

56. TikTok outliers, minus the adsPaste into Claude Cowork

For these TikTok profiles: [handles], pull the last 30 posts each with clockworks/tiktok-profile-scraper.
Check inputs and price with fetch-actor-details first and wait for my OK.
Score views against each profile's median. Flag posts tagged #ad, and posts whose like rate is under a third of the profile's median like rate, as likely paid reach.
Return the organic outliers first.

Full workflow: find outlier videos in your niche

57. How the outliers openPaste into Claude Cowork

Pull transcripts for these outlier videos: [URLs] with codepoetry/youtube-transcript-ai-scraper.
Check inputs and price with fetch-actor-details first and wait for my OK.
For each, quote the first 30 seconds and describe the hook (question, promise, conflict, demo). Say honestly when the opening does not explain the result and the title or timing is the more likely reason.

Full workflow: find outlier videos in your niche

58. Idea bank from the outliersPaste into Claude

Here are my niche's outlier videos with scores and opening hooks: [paste].
Turn them into an idea bank of 15 ideas for my channel about [topic]: working title, hook line, format, and which outlier it is modelled on.
Do not copy titles. Do not run any Apify Actors.

Full workflow: find outlier videos in your niche

59. Ideas that broke out on both platformsPaste into Claude Cowork

Here are this quarter's YouTube outliers: [paste] and TikTok outliers: [paste] for the same creators.
Match ideas that appear on both platforms (same topic or hook, even with different titles), note which came first and the gap in days.
List the ideas that have only broken out on one platform so far as candidates to test on the other.

Full workflow: find outlier videos in your niche

60. Ask a creator's whole back cataloguePaste into Claude Cowork

Using the chunked transcripts of [channel] in [folder], answer: "[question]".
Answer only from the transcripts. Cite the video title and timestamp for every claim. If the channel never covers it, say so.

Full workflow: chat with any YouTube channel in Claude

Bonus: Claude Code power users

Audit my MCP contextPaste into Claude Code

Run claude mcp list and ask me to paste my /context output.
Table each MCP server's tool and instruction tokens. Ask me, one server at a time, how often I use it.
Recommend keep, scope with a tool list, move to project scope, or disable. Show the exact config change for each and apply nothing until I say yes.

Full workflow: MCP context cost in Claude Code, measured

Research with a spend capPaste into Claude Code

Research "[question]". Use free WebSearch and WebFetch first. Only if they fail, use apify/rag-web-browser through call-actor with a maximum total charge of $0.25, and ask me before the run.
Cite a source URL for every claim and save the notes to research/[topic].md after I confirm.

Full workflow: Claude Code web research setup

Nightly changelog digest on a serverPaste into Claude Code

Set up a nightly job on this server that runs claude -p to summarise what changed overnight on these release pages: [URLs].
Show me each file before writing it, keep secrets in a chmod 600 env file with placeholders, do one dry run with --max-budget-usd 0.50, and add the crontab line only after I approve.

Full workflow: Claude Code headless mode and cron

FAQ

Frequently Asked Questions

The prompts that only analyse data you paste in work in any Claude chat. The ones that pull live data need the Apify connector or MCP server, because Claude cannot scrape Amazon reviews, Google Maps or job boards on its own. Setup takes a few minutes and is linked at the top of this page.

Most runs cost a few cents in Apify usage, and the free Apify plan includes monthly credit. Every prompt makes Claude read the Actor's live price and wait for your OK before spending, and the first run is a small test.

Custom connectors and Cowork scheduled tasks need a paid Claude plan. Claude Code needs Pro, Max, Team or Enterprise, or a Console account. The individual guides list the exact requirements for each workflow.

Yes. Prompts marked daily, weekly or monthly work as Claude Cowork scheduled tasks once a manual run looks right. Prompts that need files on your computer only run while it is on.