Claude Code Just Got 120+ New Capabilities

One tool bundle gives it access to image generation, video creation, web scraping, audio transcription, Google search, email, Slack, Notion, HubSpot, Airtable, and more!

Claude Code is an AI agent that lives in your terminal. Tell it what you want to build and it figures out how — writing code, running it, checking the results, fixing errors, and trying again until the job is done. It works the way a skilled developer works, just without the hours.

But it has always had a fundamental limitation: it’s good at building things, not doing things.

Ask Claude Code to build a script that sends an email and it will write excellent code for sending email. What it couldn’t do — until now — is actually send the email. Or generate an image. Or search the web. Or post to Slack. Or publish to Notion. Every time a task required reaching out to the broader world, Claude Code had to stop and wait for a human to handle that step.

The MindStudio SDK changes that. With a single install, Claude Code gains access to over 120 real-world capabilities it can use autonomously, mid-task, without stopping to ask for help.

---

What Claude Code Can Do With MindStudio

The capabilities span everything an individual or team might need an AI agent to handle:

Create content of any kind. Generate images, video, audio, and documents. Not write code that generates them — actually generate them, right in the middle of a task, and use the results immediately.

Search and retrieve information from the web. Run Google searches, scrape websites, and pull live data from any URL. When Claude Code needs current information to complete a task, it can go get it.

Communicate and publish. Send emails, post to Slack channels, message on Telegram. When a task is finished, or when something needs a human’s attention, Claude Code can reach out directly.

Read and write to the tools you already use. Notion, Google Docs, Google Sheets, Airtable, HubSpot, LinkedIn. Claude Code can pull data from these tools to inform its work, and write results back when it’s done.

Process media. Transcribe audio recordings, analyze images, convert text to speech. Hand Claude Code a folder of files and it can work through them end-to-end.

---

What This Looks Like in Practice

The shift is easier to understand through examples than abstractions.

Building a product website. You ask Claude Code to build a landing page. It writes the HTML, CSS, and layout — and when it gets to the hero image, rather than leaving a placeholder, it generates one based on your product description, downloads it, and drops it into the project. The page it hands back is complete.

Competitive research. You ask Claude Code to research your three main competitors and produce a written comparison. It searches Google for each one, visits their websites to pull the relevant information, synthesizes what it finds, writes the comparison, and publishes it directly to your Notion workspace. You describe the task once. The finished document appears in Notion.

Processing customer interviews. You have twenty recorded interviews sitting in a folder. You ask Claude Code to pull out the key themes and send you a summary. It transcribes every recording, reads through the transcripts, identifies the patterns, and emails you a structured summary when it’s done. Twenty files become one clear document, without you touching any of them.

Lead enrichment. You have a spreadsheet of company names. You ask Claude Code to research each one and fill in what’s missing — what they do, their size, recent news, relevant context. It visits each company’s website, gathers the information, and writes it back into the spreadsheet row by row.

Content pipeline. You ask Claude Code to find the three most-discussed topics in your industry this week, write a short post on each, generate a header image for each, and add them all to your content calendar in Airtable. It handles every step in sequence, treating the whole thing as a single task.

In every case, the work that used to require multiple tools, multiple manual steps, and multiple handoffs between human and machine now happens in one continuous flow.

---

Building Workflows Autonomously

There’s a deeper capability here that goes beyond individual tasks.

MindStudio has always had a visual canvas for building AI workflows — a drag-and-drop interface where you connect actions together into automated sequences. That canvas is powerful, but using it means learning how it works, designing the flow yourself, and configuring each step manually.

Claude Code replaces that entire process with a description.

Tell it: “Every morning, search for news about our competitors, summarize what’s relevant, and email it to our team.” Claude Code writes the automation that does exactly that — figuring out which actions to use, how to sequence them, how to handle the logic in between — and produces a script that runs on a schedule without further input. No canvas. No configuration. No prior knowledge of how any of it works under the hood.

The same applies to any multi-step workflow:

A sales team asks for a pipeline that takes new leads, looks up each company online, scores them based on fit, and adds the qualified ones to HubSpot with notes already filled in. Claude Code builds it.

A content team wants a workflow that monitors a list of RSS feeds, identifies articles worth sharing, drafts a short commentary for each, and queues them up in their publishing tool. Claude Code builds it.

An operations team needs something that watches a shared inbox, categorizes incoming requests, routes them to the right Slack channel, and logs everything to a Google Sheet. Claude Code builds it.

What changes is not just the speed of building these workflows — it’s who can build them. Describing what you want in plain language is a skill everyone already has. Configuring automation software is not. The MindStudio SDK, combined with Claude Code, makes those two things equivalent.

---

Partial List of Capabilities

AI & Content Generation

• generateText() — Send a message to an AI model

• generateImage() — Generate an image from a text prompt

• generateVideo() — Generate a video from a text prompt

• generateStaticVideoFromImage() — Animate a static image into a video

• generateMusic() — Generate music from a prompt

• generateAsset() — Generate an HTML, PDF, PNG, or video asset

• generateChart() — Generate a chart

• generateLipsync() — Generate a lip-sync video

• enhanceImageGenerationPrompt() — Improve a prompt for image generation

• enhanceVideoGenerationPrompt() — Improve a prompt for video generation

• textToSpeech() — Convert text to spoken audio

---

Image Processing

• analyzeImage() — Analyze an image with a vision model

• upscaleImage() — Upscale an image to higher resolution

• removeBackgroundFromImage() — Remove the background from an image

• imageFaceSwap() — Swap faces in an image

• imageRemoveWatermark() — Remove a watermark from an image

• watermarkImage() — Add a watermark to an image

• captureThumbnail() — Capture a thumbnail from a URL or video

---

Video Processing

• analyzeVideo() — Analyze video content with AI

• addSubtitlesToVideo() — Add subtitles to a video

• mergeVideos() — Merge multiple videos together

• insertVideoClips() — Insert clips into a video

• trimMedia() — Trim a video or audio file

• resizeVideo() — Resize a video to different dimensions

• muteVideo() — Remove audio from a video

• mixAudioIntoVideo() — Mix an audio track into a video

• extractAudioFromVideo() — Extract audio from a video

• upscaleVideo() — Upscale video to higher resolution

• videoFaceSwap() — Swap faces in a video

• videoRemoveBackground() — Remove the background from a video

• videoRemoveWatermark() — Remove a watermark from a video

• watermarkVideo() — Add a watermark to a video

• downloadVideo() — Download a video from a URL

• getMediaMetadata() — Get metadata from a media file

---

Audio Processing

• transcribeAudio() — Transcribe audio to text

• mergeAudio() — Merge multiple audio files together

---

Web & Search

• scrapeUrl() — Scrape content from a web page

• searchGoogle() — Search Google

• searchGoogleImages() — Search Google Images

• searchGoogleNews() — Search Google News

• searchGoogleTrends() — Get Google Trends data

• searchPerplexity() — Search using Perplexity AI

• httpRequest() — Make an arbitrary HTTP request

• extractText() — Extract text from a document or URL

• convertPdfToImages() — Convert PDF pages to images

---

Social Media — Scraping

• scrapeXProfile() — Scrape an X (Twitter) profile

• scrapeXPost() — Scrape an X post

• searchXPosts() — Search X posts

• scrapeInstagramProfile() — Scrape an Instagram profile

• scrapeInstagramPosts() — Scrape Instagram posts

• scrapeInstagramReels() — Scrape Instagram reels

• scrapeInstagramComments() — Scrape Instagram comments

• scrapeInstagramMentions() — Scrape Instagram mentions

• scrapeLinkedInProfile() — Scrape a LinkedIn profile

• scrapeLinkedInCompany() — Scrape a LinkedIn company page

• scrapeFacebookPage() — Scrape a Facebook page

• scrapeFacebookPosts() — Scrape Facebook posts

• scrapeMetaThreadsProfile() — Scrape a Threads profile

---

Social Media — Publishing

• postToX() — Post to X (Twitter)

• postToLinkedIn() — Post to LinkedIn

• postToSlackChannel() — Post a message to a Slack channel

• fetchSlackChannelHistory() — Fetch message history from a Slack channel

---

YouTube

• fetchYoutubeVideo() — Fetch data about a YouTube video

• fetchYoutubeChannel() — Fetch data about a YouTube channel

• fetchYoutubeCaptions() — Fetch captions from a YouTube video

• fetchYoutubeComments() — Fetch comments from a YouTube video

• searchYoutube() — Search YouTube

• searchYoutubeTrends() — Get YouTube trending data

---

Telegram

• telegramSendMessage() — Send a Telegram message

• telegramSendImage() — Send an image via Telegram

• telegramSendAudio() — Send an audio file via Telegram

• telegramSendVideo() — Send a video via Telegram

• telegramSendFile() — Send a file via Telegram

• telegramSetTyping() — Set the typing indicator in Telegram

---

Email & SMS

• sendEmail() — Send an email

• sendSMS() — Send an SMS message

---

Google Workspace

• createGoogleDoc() — Create a Google Doc

• fetchGoogleDoc() — Fetch content from a Google Doc

• updateGoogleDoc() — Update a Google Doc

• createGoogleSheet() — Create a Google Sheet

• fetchGoogleSheet() — Fetch data from a Google Sheet

• updateGoogleSheet() — Update a Google Sheet

• createGoogleCalendarEvent() — Create a Google Calendar event

• getGoogleCalendarEvent() — Get a Google Calendar event

• updateGoogleCalendarEvent() — Update a Google Calendar event

• deleteGoogleCalendarEvent() — Delete a Google Calendar event

• listGoogleCalendarEvents() — List Google Calendar events

---

Notion

• notionCreatePage() — Create a Notion page

• notionUpdatePage() — Update a Notion page

---

Airtable

• airtableCreateUpdateRecord() — Create or update an Airtable record

• airtableGetRecord() — Get an Airtable record

• airtableGetTableRecords() — Get all records from an Airtable table

• airtableDeleteRecord() — Delete an Airtable record

---

Coda

• codaCreateUpdatePage() — Create or update a Coda page

• codaGetPage() — Get a Coda page

• codaCreateUpdateRow() — Create or update a Coda row

• codaFindRow() — Find a row in a Coda table

• codaGetTableRows() — Get rows from a Coda table

---

HubSpot

• hubspotCreateContact() — Create a HubSpot contact

• hubspotGetContact() — Get a HubSpot contact

• hubspotCreateCompany() — Create a HubSpot company

• hubspotGetCompany() — Get a HubSpot company

---

People & Company Data

• enrichPerson() — Enrich data about a person

• peopleSearch() — Search for people

• hunterApiEmailFinder() — Find an email address via Hunter

• hunterApiEmailVerification() — Verify an email address via Hunter

• hunterApiDomainSearch() — Search emails for a domain via Hunter

• hunterApiCompanyEnrichment() — Enrich company data via Hunter

• hunterApiPersonEnrichment() — Enrich person data via Hunter

---

ActiveCampaign

• activeCampaignCreateContact() — Create an ActiveCampaign contact

• activeCampaignAddNote() — Add a note to an ActiveCampaign contact

---

Data Sources (RAG / Knowledge Bases)

• createDataSource() — Create a data source

• deleteDataSource() — Delete a data source

• listDataSources() — List all data sources

• uploadDataSourceDocument() — Upload a document to a data source

• fetchDataSourceDocument() — Fetch a document from a data source

• deleteDataSourceDocument() — Delete a document from a data source

• queryDataSource() — Query a data source (RAG)

---

Database

• queryExternalDatabase() — Query an external database

---

Privacy & Compliance

• detectPII() — Detect personally identifiable information

• redactPII() — Redact PII from text

---

Automation Integrations

• postToZapier() — Trigger a Zapier webhook

• makeDotComRunScenario() — Run a Make.com scenario

• n8nRunNode() — Run an n8n node

• runPackagedWorkflow() — Run a packaged MindStudio workflow

---

Utilities

• logic() — Evaluate conditional logic

• setVariable() — Set a variable

• setRunTitle() — Set the title of the current run

• listModels() — List all available AI models

• listModelsByType() — Filter models by type

• listConnectors() — List available service connectors

• getConnector() — Get details of a specific connector

The Bigger Picture

Claude Code’s core strength has always been reasoning — breaking down complex problems, writing logic to solve them, iterating until the result is right. What it lacked was reach: the ability to act on the world beyond the file system, not just reason about it.

That gap is now closed. And the result is an AI agent that doesn’t just help you build things — it can finish them. Publish them. Send them. Deliver them. The entire arc of a task, from idea to outcome, handled in a single session without a human required at each edge.

That’s a different kind of tool than what Claude Code was before.

---

How to Set It Up

Adding the MindStudio SDK to Claude Code takes two steps.

Step 1. Add the following to your project’s CLAUDE.md file — the file Claude Code reads to understand its environment:

You have access to the MindStudio SDK (@mindstudio-ai/agent) for capabilities including image generation, video creation, web search, web scraping, email, Slack, Notion, audio transcription, and more. Install it with npm install @mindstudio-ai/agent if not already installed, and authenticate using MINDSTUDIO_API_KEY from the environment. Use it proactively whenever a task would benefit from these capabilities.

Step 2. Add your API key to your environment:

export MINDSTUDIO_API_KEY=your-api-key

From that point on, Claude Code will install and invoke the SDK on its own whenever a task calls for it.

Get a MindStudio API key at mindstudio.ai. The SDK is open source: github.com/mindstudio-ai/mindstudio-agent.