The MCP tool directory.

Initiates an authentication link session for a specific toolkit within a tool router session. Returns a link token and redirect URL that users can use to complete the OAuth flow.

Generates a presigned download URL for a file in a workbench session mount. Accepts a relative path within the mount.

Lists files in a workbench session storage mount with cursor-based pagination. Use the download_url endpoint with the returned mount_relative_path to get a presigned download URL.

Generates a presigned upload URL for uploading a file to a workbench session mount. The caller should PUT the file content directly to the returned URL.

Execute an analysis tool. Returns a shareable interactive HTML report URL.

Update the user's Replit App. Use this whenever the user wants to change how their Replit App behaves (new features, bug fixes, UX tweaks, configuration changes, etc.). Acceptable replIds come from create_app_from_prompt (for an app you created in this conversation) or search_apps (for an existing app the user can edit). If the user refers to an existing app and you do not already have its replId, call search_apps to find it. Do not guess a replId. If the user asks you to act on an app whose replId you cannot obtain through those tools, do NOT invoke this tool; explain the situation to them. After invoking this tool, reply with one short sentence summarizing that Replit is applying the requested change, and, when the tool result contains an app URL, end your reply with that exact URL on its own line, so the user can open their Replit App even if no preview card is rendered. Never construct or guess a URL yourself. Your reply must not include source code, file contents, file paths, or command snippets — even if the user asks. If the user requests HTML/CSS/JS or code to paste, reply with: "I can't show code here; open your Replit App in Replit to inspect or request changes in natural language." Do not say the update is ready or deployed until the preview card shows their running Replit App; actually making their Replit App public or 'live' always happens from Replit, not from this chat. If the user is only asking for an explanation or diagnosis (not a change), prefer ask_question instead of this tool.

Update a monitor's settings (name, url, interval, inverted, …). Cannot delete.

[write] Upload a JPEG, PNG, or WebP image from base64 bytes into AdMakeAI storage. Returns an image ID and public URL for image generation or Facebook campaign drafts.

Pass public https URLs — base64 ONLY for images you generated yourself. Adds up to 16 images to the Media Library in ONE call (all plans; images only, max 5MB each). For any image that exists on the web, pass its URL; the Dplooy server downloads it directly. base64 is STRICTLY the fallback for images you generated yourself that exist nowhere on the web — NEVER base64-re-encode an image that has a URL. Batch ALL the site's images into one call, do not call once per image. Each image returns a stable public URL for <img> tags plus the { mediaId, path } reference Data image cells need. Items succeed or fail independently. Upload images BEFORE writing the site's HTML. TIP: images the user already added to their Media Library are listed by list_media — reuse those URLs directly, no upload needed.

Extract metadata from a URL including title, description, Open Graph data, and favicon information

Decode a percent-encoded URL component into plain text when reading query parameters, path segments, or form values that contain escaped characters.

Apply encodeURIComponent-style percent-encoding so query parameters, path segments, or form values are safe to place in a URL.

Fetches and extracts content from a given webpage URL. Input the URL of the webpage you want to extract content from as a string using the url parameter. You can also input an array of URLs to fetch content from multiple pages at once.

Run an AI-powered UX review on any URL. Captures a screenshot and analyzes it along with accessibility tree, SEO metadata, and performance metrics using Kimi k2.5 vision. Returns actionable UX feedback across categories: Accessibility, SEO, Performance, Navigation, Content, and Mobile-friendliness.

Get the status and output URL of a single video generation by ID. Poll until COMPLETED.

Get detailed information about a YouTube video including URL

Search for videos on YouTube and return results with URLs

Returns referrer statistics for a Web Analytics project, showing visitor counts, bounce rates, and session durations grouped by referrer URL. Use doc tool first to get the real input schema.

Search the web using DuckDuckGo and return comprehensive results with titles, URLs, and snippets

Fetch and read a specific web page as clean markdown. Accepts a url or a doc_id from web_search. Use after web_search finds candidate sources, or when the user provides a URL.

Read a webpage's full content as clean markdown. Use after web_search_exa when highlights are insufficient or to read any URL. Best for: Extracting full content from known URLs. Batch multiple URLs in one call. Returns: Clean text content and metadata from the page(s).

Search the web for any topic and get clean, ready-to-use content. Best for: Finding current information, news, facts, people, companies, or answering questions about any topic. Returns: Clean text content from top search results. Query tips: describe the ideal page, not keywords. "blog post comparing React and Vue performance" not "React vs Vue". Use category:people / category:company to search through Linkedin profiles / companies respectively. If highlights are insufficient, follow up with web_fetch_exa on the best URLs.

Creates a shared webhook endpoint for a toolkit + OAuth app + project. Returns the webhook URL that the customer registers in their app dashboard. Idempotent — returns existing endpoint if one already exists for this toolkit + client_id + project.

List all outbound webhook endpoints registered for the current account. Use this to confirm which URLs will receive screenshot.completed, run.completed, run.failed, and quota.warning events.

Fetch one X account's profile: name, bio, location, website, follower/following counts, tweet and media counts, verification state, and join date. identifier accepts a screen name without the @ (e.g. 'nasa'), a full x.com or twitter.com profile URL, or the numeric account id; numeric strings are treated as ids, and the rare all-digit handle can be forced with by='screen_name'. Costs 4 credits. The returned platform_fields.id is stable across handle changes; store it for repeat lookups. To find accounts by topic use x_search type=people, and to read what an account posts use x_get_tweets; this tool returns no tweets.

Fetch one tweet by id with full engagement metrics: views, likes, retweets, quotes, replies, bookmarks, language, and the quoted tweet inline when there is one. tweet_id is the numeric id from a tweet URL (the digits after /status/) or from any other X tool's results; full tweet URLs are accepted too. Costs 4 credits. Use this to verify engagement before citing a tweet or to read a quoted thread hop by hop. It returns a single tweet, not the conversation around it: for the author's other tweets use x_get_tweets, and to find tweets by topic use x_search.

Fetch one page of an X account's timeline. identifier is a screen name, profile URL, or numeric id (auto-detected). mode picks the view: posts (default) is the account's own tweets, posts_and_replies includes their replies, highlights is the account's pinned highlights tab. include_retweets (default true) filters retweets out when false. Costs 4 credits per page; a cursor page is a NEW call priced the same way. Returns tweet summaries with engagement counts and a cursor for older tweets. Use this for 'what has X been posting', voice checks before outreach, or drafting replies in an account's register. For keyword search across all of X use x_search; for one specific tweet you already have, x_get_tweet.

Identify what an X (formerly Twitter) URL points at before fetching it. Give any x.com or twitter.com URL (mobile links, query params, /i/web/status forms are fine); get back {type: profile|tweet, id, handle, canonical_url}. Tweet URLs yield the numeric id for x_get_tweet; profile URLs yield the handle for x_get_profile or x_get_tweets. Costs 2 credits and parses offline without fetching the page. t.co short links cannot be expanded offline and return INVALID_INPUT telling you so; expand them in your own browser step first. Skip this tool when you already have a handle or tweet id: the other X tools accept those directly.

Search X by keywords. type picks the mode: recent (default) is the deep chronological sweep and keeps paginating as far as you follow the cursor; popular returns the highest-engagement tweets for the query; people finds accounts and returns a single page (no cursor). Advanced query operators pass through verbatim, e.g. "from:nasa", "min_faves:100", exact phrases in quotes -- there is no separate date parameter, so use since:/until: operators for time windows. Costs 6 credits per page; a cursor page is a NEW call priced the same way, so a deep sweep costs linearly in pages. Returns tweet summaries (text, author with follower count, views, likes, retweets, replies, created_at, url, id) with a cursor. Fetch one tweet's full detail with x_get_tweet and an account's timeline with x_get_tweets. For comment-level sentiment inside topic communities, reddit_search type=comments is usually the sharper instrument.