Overview Claude ChatGPT Other MCP clients Tool reference Sources Credits & limits Good to know

MCP Docs

Setup, tools, and what
each call costs.

The reference for the hosted Adlicio MCP connector: how to add it to your AI app, every tool it exposes with its parameters, and how the metering works. New here? The usage guide covers what to type once you are connected.

Overview

One endpoint, no API keys.

Adlicio is a hosted remote MCP server. Any client that speaks MCP over HTTP can connect; auth is a normal sign-in with your Adlicio account, so there is nothing to generate, paste or rotate.

Endpointhttps://mcp.tryadlicio.com/mcp

TransportStreamable HTTP

AuthOAuth 2.1 with PKCE. Sign in with your Adlicio account when the client asks.

PrerequisiteAn Adlicio account. Free covers a Reddit taste; Pro and All Access lift the caps.

Local MCP

Claude Desktop's local stdio MCP runs through the commentscraper CLI instead (All Access, four tools). The developer docs cover the CLI, the local config, the REST API and webhooks.

Setup · Claude

One click, or six taps.

Claude web, mobile, Cowork and Desktop all use the same hosted connector. The fast path pre-fills everything:

Remote connector URL

https://mcp.tryadlicio.com/mcp

Add to Claude

Claude web, mobile, Cowork

01Open Claude and go to Customize > Connectors.
02Click +, then choose Add custom connector.
03Name it Adlicio and paste https://mcp.tryadlicio.com/mcp.
04Click Add, then Connect.
05Sign in with the Google account you use for Adlicio or the Reddit Comment Scraper extension.
06In a chat, open + > Connectors and enable Adlicio for that conversation.

Team and Enterprise

01An Owner adds the connector in Organization settings > Connectors.
02Choose Custom > Web and paste the same remote MCP URL.
03Each team member then connects their own Adlicio account from Customize > Connectors.

Open Claude admin setup

For Claude Code, one command registers the server; run /mcp afterwards to sign in.

Claude Code

claude mcp add --transport http adlicio https://mcp.tryadlicio.com/mcp

Setup · ChatGPT

Add it as a connector app.

ChatGPT web and desktop

01Open Settings, then Apps & Connectors.
02Under Advanced settings, turn on Developer mode (needs a paid ChatGPT plan).
03Choose Create, name it Adlicio, and paste https://mcp.tryadlicio.com/mcp.
04Pick OAuth as the authentication and save.
05Sign in with your Adlicio account when the browser window opens.
06In a chat, enable Adlicio from the + menu, then ask your first research question.

Tool names differ slightly

ChatGPT shows the same tools with shorter descriptions than Claude does. Behavior, parameters and costs are identical.

Setup · Everything else

Any client that speaks remote MCP.

Perplexity, Grok and Le Chat each have a connector setting: paste the endpoint URL, keep OAuth as the auth method, and sign in when prompted. For Cursor and other editors that read an mcp.json, this is the whole config:

mcp.json

{
  "mcpServers": {
    "adlicio": {
      "url": "https://mcp.tryadlicio.com/mcp"
    }
  }
}

The server handles dynamic client registration, so there is no client id to request. If your client only supports stdio servers, use the local CLI route in the developer docs.

Tool reference

Sixteen tools, two jobs.

Your AI discovers all of these on its own; you ask in plain language. Half collect and rank customer language, half read and write the workspace you already have. Optional parameters carry a trailing ?.

Research

research_topic

1 credit

Every plan

Search Reddit and scrape the matching threads in one call. The starting point for ranked angles and hooks.

keyword, subreddits?, limit?, exclude_scraped?, mode?, max_comments?, min_score?, max_depth?, sort?

search_reddit

Free

Every plan

Find the threads most relevant to a keyword: URLs, titles and snippets, no scraping. Threads you already scraped come back flagged, so you can reread them free.

keyword, subreddits?, limit?

scrape_url

0 to 2

Every plan*

Pull the comments or reviews from one supported URL. Most sources are free; ten premium sources use 1 credit and a live Amazon scrape uses 2.

url, mode?, max_comments?, max_threads?, min_score?, max_depth?, sort?, include_body_only?

find_scroll_stoppers

3 credits

All Access

The top-performing public short-form videos for a niche or competitor, each with its on-screen hook and engagement, ranked best first. TikTok and Instagram.

keyword, platform?, limit?

find_competitor_ads

Free

All Access

The ads a brand is running right now on Facebook and Instagram, from the public Meta Ad Library: the live copy, hooks and offers.

query, country?, active_status?, limit?

find_products

1 credit

All Access

Search Google Shopping listings: title, price, old price when discounted, rating, review count, merchant and URL. Listings, not review text.

query, max_results?

start_deep_scrape

3 credits

All Access

Start a background scrape of all available comments on one YouTube video, up to 20,000. A live job for the same URL is reused without another charge.

url, max_comments?

get_deep_scrape

Free

All Access

Check the status of a deep scrape job and read the result when it lands.

job_id

* scrape_url works on Reddit URLs from every plan; all other sources need All Access.

Workspace

whoami

Free

Every plan

The connected account, its plan, and the research credits remaining this cycle.

none

list_history

Free

All Access

List the scrapes and researches this account already ran, from the dashboard, extension, CLI and chat, newest first.

query?, platform?, limit?, offset?

get_history_item

Free

All Access

Reread the stored comments of one past scrape, without scraping anything again.

id, max_comments?, offset?

search_history

Free

All Access

Search every comment the account has ever collected, across all scrapes and platforms, for specific customer language.

query, platform?, limit?

get_brand_research

Free

All Access

Read the avatars, ad angles, verbatim quotes and saved scripts a brand already has. Call it with no arguments to list the account's brands.

brand?

create_brand

Free

All Access

Create a brand workspace, the same brands shown in the Adlicio dashboard. An exact name that already exists returns the existing brand.

name, url?, niche?, audience?, offer?, key_claim?, tone?, mechanism?

start_brand_research

3 credits

All Access

Kick off the full brand research pipeline in the background, exactly like the dashboard's Run research button. Poll get_brand_research for the result.

brand

save_script

Free

All Access

Save a finished ad script into a brand's Scripts library, where the dashboard shows it.

brand, script, title?

Sources

Where scrape_url can read.

Point it at a public thread, video, product page or review page on any of these. Reddit is included on every plan; the rest unlock with All Access.

No credits

RedditYouTubeApp StoreGoogle PlaySteamTrustpilotReviews.ioG2AliExpressHacker NewsQuoraProduct HuntStack ExchangeDiscourse forumsphpBB forumsXenForo forumsvBulletin forumsInvision forumsShopify product pagesDisqus commentsNotion pages

1 credit per delivered scrape

FacebookInstagramTikTokEtsyWalmarteBayBetter Business BureauSephoraUltaTripAdvisor

2 credits per live scrape

Amazon (live reviews)

Amazon only charges 2 credits when a fresh live read actually runs; cached and archived results are free.

Credits & limits

One meter: research credits.

Searching and reading are free. Credits are spent when Adlicio collects or ranks something new, and only after the result is actually delivered. whoami shows your balance any time.

Free6 research credits to start3 Reddit pulls in chat, 3 Reddit searches a day

Pro30 research credits a month10 Reddit pulls a month in chat, 10 Reddit searches a day

All Access250 research credits a monthEvery tool and every source, plus the CLI

Team750 research credits a month, sharedEverything in All Access across 4 seats

All Access also carries generous daily fair-use ceilings (200 Reddit searches, 150 scrapes) that normal research never touches. Full pricing lives on the pricing page.

Good to know

The fine print, minus the fine.

Repeats are not re-chargedScraping the same URL again within 15 minutes does not use a second credit.

Reread before you re-scrapesearch_reddit flags threads already in your history with their scrape id. Reading them back through get_history_item is free.

One credit poolResearch credits are one shared monthly pool across the dashboard, the Chrome extension, the CLI and chat. They reset with your billing cycle.

Free means zero creditsA tool marked free uses no credits, but most tools beyond the Reddit set still need an All Access plan. The plan column in the reference is the one that gates access.

Out of credits is not an errorWhen a call needs more plan than you have, the chat gets a clear upgrade prompt instead of a failure.

Tools stopped showing upStale tool lists live in the AI app, not in Adlicio. Disconnect and reconnect the connector in that app's settings.

Instagram runs in the backgroundComment collection on Instagram posts and reels can take a moment. If the first call returns pending, retry the same URL.

Disconnect the connector anytime from your AI app's settings; revoke authorized devices from your account.

Connect once. Ask in plain language.

Three runs are free, and the first ranked answer usually pays for the setup time on its own.

Add Adlicio to Claude Read the usage guide