Instagram + YouTube Data for AI Agents (Yugenox)
io.github.ArpitGandhi1934Published by GitHub account ArpitGandhi1934v1.2.0Updated Oct 30, 2026
Public Instagram and YouTube data for AI agents: reels, videos, channels, comments. Pay per result.
Overview
Apify-hosted MCP server that lets an assistant scrape public Instagram and YouTube data — reels, videos, channels, comments — billed per result.
- What you can do
- Run the YouTube scraper for video stats, top comments and a dislike estimate from search terms, channels, playlists or video URLs
- Run the Instagram scraper for public posts, reels, profiles, comments and reel transcripts, with engagement and partnership signals
- Read an Actor's README and current price before starting a paid run
- Inspect, abort and retrieve results of runs through Apify's run helpers (run status, dataset items, key-value records)
- Pick narrower pinned variants for Instagram comments, reel transcripts or YouTube AI summaries
- When to use it
- Useful when an assistant needs public social-media data for research, content analysis, comment mining or creator vetting without a YouTube API key or Instagram login. It fits agents that can pay per result through an Apify account.
- Requirements
- Remote streamable HTTP endpoint at mcp.apify.com with a URL that pins the desired Actors. An Apify account is required: OAuth sign-in on first tool call, or an Authorization Bearer header carrying your own Apify token for clients without remote OAuth. Clients that only run local servers need the mcp-remote bridge via npx.
Installation
In SourceWeft
- Open Instagram + YouTube Data for AI Agents (Yugenox) in the dashboard and add it to a workspace.
- Connect the required credentials. They are stored privately and never shown on public pages.
- Enable the server for the chats that should use its tools.
Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.
Use with an agent
Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.
Read https://sourceweft.com/mcp/io.github.ArpitGandhi1934%2Fyugenox-social-data/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.Other MCP clients
Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.
{
"mcpServers": {
"instagram-youtube-data-for-ai-agents-yug": {
"type": "http",
"url": "https://mcp.apify.com/?tools=fetch-actor-details,yugenox/instagram-scraper,yugenox/youtube-scraper"
}
}
}README
Instagram + YouTube data for AI agents (MCP)
Connect two public-data scrapers to ChatGPT, Claude, Claude Code, Cursor, VS Code, Windsurf, n8n or your own agent in about a minute. No YouTube API key, no Instagram account, no shared keys: you sign in to your own Apify account with OAuth and pay per result.
Both are published by Yugenox Corporation on the Apify Store. This repo holds no server code:
Apify hosts the MCP server at mcp.apify.com, and the URLs below pin it to these two Actors.
Step-by-step guides for these Actors — pricing worked out, input recipes, honest limits and side-by-side comparisons — live at yugenox-data.vercel.app, including how to use them in ChatGPT, Claude, Claude Code and Cursor.
Pick a server
Each server exposes its pinned Actors as tools (for example yugenox--youtube-scraper), plus
fetch-actor-details (lets the agent read the README and current price before a paid run)
and Apify's run helpers get-actor-run, get-dataset-items, get-key-value-store-record
and abort-actor-run.
Also in this repo: three single-purpose servers for Instagram comments, Instagram reel transcripts and YouTube AI summaries (see Single-purpose servers), and four for Canadian grocery prices, Canadian retail store stock, Realtor.ca + Kijiji, and company career-site jobs (see More pinned servers). They connect the same way: swap in their name and URL in any of the steps below.
Sign-in: the first tool call opens an Apify sign-in page (OAuth). Runs bill to the account you sign in with. Never paste a token into one of these URLs.
Connect in 60 seconds
Claude Code
Then run /mcp, pick yugenox-social and authenticate in the browser.
Or install the plugin, which adds the same server plus two skills that state the cost before each run:
The plugin folder, plugins/yugenox-social-data, also follows the
Agent Plugins 1.0 format (plugin.json, mcp.json, skills/), so
other clients that load Agent Plugins can install the same server and skills from it.
Claude.ai and Claude Desktop
Settings → Connectors → Add custom connector. Name it yugenox-social, paste the server
URL, click Add, then Connect and sign in to Apify. Leave the OAuth client fields empty;
Apify registers the client automatically.
ChatGPT
ChatGPT on the web → Settings → Apps & Connectors → Advanced settings → turn on Developer mode. Back in Apps & Connectors, click Create, paste the server URL, set Authentication to OAuth, confirm, and sign in to Apify. In a chat, enable the connector from the + menu under Developer mode. (Menu labels move between ChatGPT releases; the flow stays the same.)
Cursor
[Add YouTube + Instagram to Cursor]
YouTube only: install · Instagram only: install
Or add it to ~/.cursor/mcp.json:
Cursor then asks you to sign in: open Settings → MCP and click the login button next to the server to finish the Apify sign-in.
VS Code (GitHub Copilot agent mode)
[Install in VS Code] [Install in VS Code Insiders]
YouTube only: install · Instagram only: install
Or add it to .vscode/mcp.json (note the servers key):
VS Code asks you to sign in to Apify when the server starts.
Windsurf
Add to ~/.codeium/windsurf/mcp_config.json, then refresh the MCP panel and sign in:
n8n
Add an AI Agent node, then attach an MCP Client Tool:
- Endpoint: the server URL
- Server Transport: HTTP Streamable
- Authentication: MCP OAuth2 where your n8n version offers it; otherwise Bearer Auth with a credential that holds your own Apify token (kept in n8n's credential store, never in the URL)
- Tools to include:
yugenox--youtube-scraper,yugenox--instagram-scraper,fetch-actor-details
Clients without remote OAuth
Any MCP client that only runs local (stdio) servers can use the mcp-remote bridge, which
opens the Apify sign-in in your browser:
Every link and config above is generated from one file: install-links.json.
Use it from code
Minimal, runnable examples in examples/. Each reads your own APIFY_TOKEN
from the environment and sends it as an Authorization header.
What it costs
You pay Apify per result, from your own account. Prices below are the Store list prices on 2026-09-26; the Actor pages are the source of truth.
- YouTube: $0.002 per video row. Top comments and the dislike estimate ride inside the
row, so 100 videos with 20 comments each is $0.20. The same job with
streamers/youtube-scraperplus its separatestreamers/youtube-comments-scraperis $4.40 at their Free-tier prices ($0.40 for videos + $4.00 for 2,000 comments). - Instagram: $0.0019 per row (post, comment, profile, place or keyword). Latest comments, Instagram's AI summary and view counts are included. Transcripts add $0.004 per started minute of audio, so a 30-second reel with its transcript is about $0.0059. Reels with no detectable speech are usually skipped for transcription.
- On Instagram, rows removed by your filters, sources that return nothing and private accounts cost nothing.
- On YouTube, every saved row is billed at $0.002, including error rows: an empty or
over-filtered search, an unknown channel or playlist, and an unavailable, private or
age-restricted video each save one row with an
errorfield explaining why.
Good habit for agents: set maxItems on every call and say the estimate before running.
The plugin skills and example prompts do this.
Example prompts
YouTube:
- "Pull the last 50 videos from @NASA and flag the ones above 2x the channel's median views."
- "Top 30 videos about 'home espresso' uploaded this month, sorted by views, with 10 top comments each. What do viewers complain about?"
- "Like-to-dislike ratio for these 40 video URLs." (Dislikes are estimates, see below.)
- "Audit @somecreator before we sponsor them: average views over the last 30 uploads and the sponsor and social links in their descriptions."
Instagram:
- "Transcribe the last 20 reels from @nasa and group the opening lines into hook types."
- "Every post [a brand handle] published in the last 30 days with likes, comments and plays."
- "Vet @somecreator: engagement rate over the last 30 posts, paid-partnership posts, and 20 similar accounts with a contact in their bio."
- "Reels using this sound: https://www.instagram.com/reels/audio/271328201351336/"
Honest limits
- Public data only. Nothing behind a sign-in: no stories, follower lists, tagged posts, private accounts, members-only videos or YouTube Studio analytics.
- YouTube dislikes are estimates from the community Return YouTube Dislike database, most accurate on older videos. YouTube removed public dislike counts in 2021.
- YouTube transcripts are not a feature here. With
includeSubtitleson, the available subtitle languages are listed (best-effort); subtitle text is not reliable. - YouTube
durationis filled on search rows only for now; channel, playlist and video-URL rows leave it empty. - Instagram hashtags and keywords return the curated top posts Instagram shows publicly, typically about 60 per term.
- Instagram creator signals are public engagement numbers. There are no audience demographics or fake-follower scores, and public reel play counts can undercount, so they are not a basis for creator payouts.
- Results can include public usernames and comment authors. Handle them under GDPR, PIPEDA or CCPA.
Single-purpose servers: Instagram comments, reel transcripts, YouTube summaries
Same model as above: Apify hosts the MCP server, the URL pins it to one Actor, you sign in to your own Apify account (OAuth) and runs bill per result to that account. Public data only.
The Actor pages are the source of truth for prices, and fetch-actor-details lets the agent read
the price in force before it starts a run.
yugenox-instagram-comments
Cursor · VS Code · VS Code Insiders
- "Every comment on these three reels that mentions "restock" or "sold out", newest first."
- "The 200 most-liked comments on this post, grouped by what people are asking for."
yugenox-reels-transcript
Cursor · VS Code · VS Code Insiders
- "Transcribe these 10 reels and give me the main claim of each in one line."
- "Which of these reels mention a discount code, and what is the code?"
yugenox-youtube-summarizer
Cursor · VS Code · VS Code Insiders
- "Summarize this 2-hour podcast with chapters and timestamp links."
- "For these 5 product review videos, what are the top 3 complaints?"
Limits for these servers. Instagram shows comment replies only to logged-in accounts, so the comments server returns top-level comments. Reel transcripts and video summaries are written by Google's Gemini and can contain mistakes: check anything important against the original. The summarizer reads public YouTube videos only. Comment and reel rows include public usernames: handle them under GDPR, PIPEDA or CCPA.
More pinned servers: Canada retail, real estate and jobs
Same model as above: Apify hosts the MCP server, the URL pins it to our Actors, you sign in to your own Apify account (OAuth) and runs bill per result to that account. Public pages only.
Prices are the main per-row Store list prices on 2026-10-06. Other row types (store stock rows,
warehouses) and optional add-ons (product, listing or job details, live stock checks) are priced
separately, and several of these Actors also charge a start fee per run, billed per GB of run
memory (at most $0.005 per GB, including the changes scheduled for 2026-10-11). The Actor pages are the source of truth, and
fetch-actor-details lets the agent read the price in force before it starts a run.
yugenox-canada-grocery
Cursor · VS Code · VS Code Insiders
- "Price 2 L milk, a dozen eggs and chicken breast at the three No Frills and three Loblaws stores nearest M5V 2T6, with PC Optimum member prices."
- "This week's flyer deals on chicken near K1P 1J1, with the price per kg."
- "Costco.ca coffee on sale right now, with the date each deal ends."
yugenox-canada-store-stock
Cursor · VS Code · VS Code Insiders
- "Which of the 5 Home Depot stores nearest Ottawa have a DeWalt 20V drill in stock, and in which aisle?"
- "Clearance items in stock at Princess Auto Mississauga, biggest discount first."
- "How many units of this LCBO product does each store have on hand?"
yugenox-realtor-ca
Cursor · VS Code · VS Code Insiders
- "Realtor.ca rental listings in Vancouver: the 2-bedroom ones sorted by price, with the listing brokerage."
- "Which brokerages have the most agents listed in North York on Realtor.ca?"
- "Honda Civics for sale on Kijiji in Calgary under $15,000, with year and kilometres."
yugenox-ats-jobs
Cursor · VS Code · VS Code Insiders
- "Remote backend engineer jobs at Stripe, Airbnb and Notion, with salary ranges where posted."
- "Engineering roles on NVIDIA's Workday career site posted in the last 7 days."
- "Nursing jobs in Texas posted this week, across the largest iCIMS employers."
Limits for these servers. Public pages only: no logins, and Realtor.ca coverage is active listings (no sold history). Prices and stock are what each retailer shows for the chosen store or postal code at run time. The Realtor.ca listings search covers 18 major cities and returns up to 500 listings per city, newest first, and the agent search does not take postal codes. Instacart and Home Depot Canada work best on Apify residential proxies; on a plan without them they fall back to datacenter IPs, which is slower and may miss some results. Salaries appear only where the employer publishes them. Realtor.ca rows include agent names and phone numbers, Kijiji listing text can include personal details, and Home Depot product details include reviewers' display names: handle them under PIPEDA, GDPR or CCPA.
Not affiliated with or endorsed by any retailer named here, Instacart, Flipp, CREA or REALTOR.ca, Kijiji, the applicant tracking systems named, or the employers and brands in the example prompts.
MCP Registry
Listed in the official MCP Registry
under the io.github.ArpitGandhi1934/ namespace:
The entries live in registry/ and are published by GitHub Actions on each
version tag.
Guides
Longer write-ups for each server — pricing math, input recipes, honest limits and comparisons — on yugenox-data.vercel.app. All public data — no Instagram login or YouTube API key needed; the Actor pages on Apify are the source of truth for prices.
- Use them in ChatGPT, Claude, Claude Code and Cursor — MCP setup guide
- YouTube — YouTube Data API alternative · a channel's videos to CSV · comments by video · compare YouTube scrapers
- Instagram — scrape without login · reels to transcripts · influencer vetting · compare Instagram scrapers
- Instagram comments — export comments · comment monitoring
- Canadian grocery and flyers — grocery prices API · weekly flyer API
- Realtor.ca — listings scraper · Realtor.ca data API
- Jobs (ATS) — Workday jobs API · Greenhouse, Lever and Ashby · compare job-posting data providers
About
Built by Yugenox Corporation, Toronto. Questions or a bug: open an issue here or on the Actor's Issues tab on Apify. Uses public data only. Not affiliated with or endorsed by YouTube, Google, Instagram, Meta, OpenAI or Anthropic.
MIT licensed (this repo's configs and examples). The Actors are paid services on Apify.
Source: README.md at commit 44000c7
Tools
0Version history
3- v1.2.0LatestOct 8, 2026
- v1.1.0Oct 6, 2026
- v1.0.0Sep 26, 2026
Score
85% of the applicable checks met
How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.
- Reached in the last 30 daysRequired25 pts
- Auth declaredRequired10 pts
- Tools listed15 pts
Answer tools/list without a credential, or ship a package that starts in a sandbox without one.
- DescriptionRequired15 pts
- Logo10 pts
- Documentation15 pts
- License10 pts
