Apify in Claude: what the connector does and how Linkeddit tells it where to look

Apify's MCP documentation says the server lets AI apps find and run Actors, access storage and results, and search Apify's own docs. Its tool groups are actors (search-actors, fetch-actor-details, call-actor), docs, runs (get-actor-run, get-actor-run-list, get-actor-run-log, abort-actor-run), storage (get-dataset-items and others), tasks and schedules. The default set also includes two Actors, apify/rag-web-browser and apify/web-fetch.

One detail trips people up. According to Apify's docs, call-actor returns run status and storage IDs, not the scraped rows, so Claude has to read the output with get-dataset-items afterwards. The Agent Index listing for the connector shows a single tool, readme, so check Apify's own docs for the live tool set rather than the directory card.

Without an Apify account Claude can use only search-actors, fetch-actor-details, search-apify-docs and fetch-apify-docs. Every other tool needs a token. Apify also states that full-permission Actors are excluded from search and execution through the server.

Add the server URL https://mcp.apify.com as a connector and approve the sign-in. Apify lists OAuth as the recommended route, a bearer token as the alternative, and a local stdio server started with npx -y @apify/actors-mcp-server when you prefer to run it yourself.

In Claude Code, Apify documents a plugin. Google's AI Overview for the query "apify claude" describes the same split: the desktop route uses the MCP server, and the terminal route uses the plugin. Add Linkeddit at https://mcp.linkeddit.com/mcp beside it, and Claude can use both in one session.

Apify scrapes whatever you point it at. It does not choose the targets. A guide on use-apify.com begins its workflow with "scope the competitor URL list", then names Actors such as Website Content Crawler and Google Search Results Scraper. The list itself is left to you.

On Linkeddit's own test project, plausible.io, the competitor footprint covers 358 domains drawn from 59 checked keywords. matomo.org ranks in the Google top 10 on 11 of them, and was cited 23 times across the AI answer engines in the latest finished run, 21 of them by ChatGPT. umami.is ranks on 7 keywords, with 17 AI answer citations. Those are the pages worth crawling first, and Linkeddit names them with evidence.

A founder describes the usual alternative this way. "What surprises me is that many companies still rely on spreadsheets, saved tabs, screenshots, and occasional alerts for competitor tracking," wrote a startup founder on Reddit. A measured competitor list replaces the saved tabs, and Apify's runs replace the screenshots.

The workflow has four steps: Linkeddit names the competitor domains and the pages that beat you, Apify's Actors collect those pages' content, Claude compares them, and you decide what to change. The full guide, with the exact prompt and a measured run, is the Apify competitor analysis page linked from this hub.

The search results for the connector-plus-competitor query show why the pairing is open. Skill directories such as mcpmarket describe Claude running Apify scrapers for pricing, reviews and engagement on platforms like Google Maps, Booking.com, Instagram and TikTok. None of them starts from the competitor list Google rank data produces.

Partly, and the data comes first. Linkeddit's Answer Radar records which pages the five engines cite when they answer your buying questions, and answer_gaps_get lists the questions where an engine recommended someone else. On the test project the top open gap is "Not represented for: 'Umami vs Matomo for self-hosted analytics'", with a confidence of 1 and an evidence score of 22.

Reading the cited pages is a separate step. answer_sources_fetch does that, and it spends credits, so Linkeddit returns an estimate first and Claude should wait for your yes. Once Claude has the cited URLs, Apify can crawl those pages, for example to collect the author and contact details a page publishes. Claude then drafts the outreach. Nothing here sends a message, and you review every draft.

Check what the run costs and what it is allowed to touch. Apify bills its Actors on its own plan, separate from Linkeddit's $60 a month, and Apify describes pay-per-event Actors that can be paid per request. Ask Claude to call fetch-actor-details first and report the Actor's inputs and pricing model before it calls call-actor.

The use-apify guide covers legal boundaries and cost estimates but, on a read of it, says little about failed runs or site layout changes. Ask Claude to read get-actor-run for the status of every run and to say so when one fails, because a half-empty dataset looks like a quiet competitor.