# Descript MCP MCP > Descript lets your AI agent manage your entire video and podcast pipeline without you ever touching the desktop app. It gives direct access to project details, full transcripts for repurposing, and export management—all conversational actions. ## Overview - **Category:** image-video - **Price:** Free - **Tags:** transcription, video-editing, podcast-production, media-assets, screen-recording, collaborative-editing ## Description You can use this MCP to handle media assets from start to finish, treating Descript like a backend service rather than a graphical editor. Need to pull a transcript for a blog post? Or check if that final 4K video render is done? Your agent handles it conversationally. You ask it to list all projects or retrieve the full text for 'Podcast Ep 47' and get an immediate, structured response. It can also track exports—checking status, listing files, and triggering new renders. For teams building multi-platform content strategies, this is huge. Instead of manually logging into Descript to pull transcripts and then jumping to a separate CMS to write show notes, your agent can automate that whole sequence. This ability to chain multiple services together—say, pulling the transcript data from here and feeding it directly into a billing MCP for client invoicing—is what makes Vinkius such a powerful platform. The entire process runs securely inside an isolated sandbox on Vinkius, so you never worry about where your keys live or how those actions are audited. You just get the result. ## Tools ### create_export Starts the process of rendering a specified Descript project into a new media file. ### create_transcription Takes an audio or video file and generates a raw text transcript from its spoken content. ### get_project Retrieves the full metadata and current status for a specific Descript project ID. ### get_transcription Fetches the already generated transcription result, useful for retrieving clean text data. ### list_drives Returns a list of all available team drives and organizational containers in Descript. ### list_exports Retrieves a list of recent or pending export jobs, including their status and file types. ### list_projects Lists all Descript projects accessible to the account, providing names and last modified dates. ### list_templates Shows a list of default or custom project templates available for new content creation. ## Prompt Examples **Prompt:** ``` List all my Descript projects. ``` **Response:** ``` I found 8 projects in your Descript drive: 'Q1 Product Demo' (last edited 2h ago), 'Podcast Ep 47' (last edited yesterday), 'Customer Interview - Acme', and 5 more. ``` **Prompt:** ``` Get the full transcript of 'Podcast Ep 47'. ``` **Response:** ``` Here's the transcript for 'Podcast Ep 47' (duration: 42:15). Speakers detected: Host (Sarah) and Guest (Mark). Full text with timestamps extracted. Would you like me to convert this into show notes or a blog post? ``` **Prompt:** ``` Check if my latest video export is ready for download. ``` **Response:** ``` Your export of 'Q1 Product Demo' is complete. Format: MP4 (1080p). File size: 245 MB. Download link is ready and valid for 24 hours. ``` ## Capabilities ### List all projects and drives Finds every project and team drive within Descript to scope out exactly what media assets exist. ### Get specific project details Pulls metadata for a single project, including its compositions and associated files. ### Extract full transcripts Generates the complete text transcript from any uploaded audio or video file. ### Manage content exports Initiates new media renders, checks the status of existing exports, and retrieves download links for final files. ### Access templates and settings Lists available project templates or provides system-level details about your Descript account structure. ## Use Cases ### The Show Notes Problem A podcast producer finishes an episode and has 45 minutes of raw audio. Instead of transcribing it manually and then copying the text into Notion, they ask their agent to use `create_transcription` on the file, getting structured, clean text ready for immediate show note generation. ### The Asset Retrieval Headache A marketing team needs a specific video asset from 'Q1 Product Demo' but doesn't know which folder it's in. They ask the agent to use `list_drives` and then narrow down with `get_project`, finding the exact file metadata instantly. ### The Follow-Up Render A client approves a video draft, but needs it in 1080p MP4 format. The user asks their agent to use `create_export` with the correct parameters and then checks the status using `list_exports`, getting a direct download link when ready. ### The Multi-Platform Content Flow A content strategist needs transcripts from three different projects. They ask the agent to use `list_projects` first, then loop through and call `get_transcription` for each one sequentially, compiling all raw text into a single document. ## Benefits - Need to repurpose a video? Instead of manual copy-pasting, your agent can use `get_transcription` to pull full text and timestamps for blog posts or social captions. - Don't waste time checking render status. You can ask the agent to check if an export is ready using `list_exports`, getting immediate confirmation and download links. - Building a content pipeline? The ability to chain together actions—like first listing projects, then extracting transcripts, and finally creating exports—automates your entire workflow. - Keep track of everything with `list_drives` and `list_projects`. You can scope the search space instantly so you know exactly which asset you're talking about. - Triggering renders is simple. Use `create_export` to kick off a new video or audio version without needing GUI access, letting your agent monitor completion. ## How It Works The bottom line is: you use your AI client to talk to Descript's API through the MCP, and it brings back organized, actionable data for whatever workflow you're running. 1. Ask your AI agent to list the primary projects you need help with, so it can scope out the assets. 2. Direct the agent to perform a specific action, like getting project metadata or pulling a transcript for a given file. 3. The agent returns structured data—be that a download link, a text block of timestamps, or a list of available files. ## Frequently Asked Questions **How do I use `list_projects` with Descript MCP?** You just ask your agent to 'List all my projects.' It scans the entire account and returns names, giving you a quick overview of everything available in your drive. **Can I use `create_transcription` on different file types?** Yes. The tool supports both audio files and video files. You just need to provide the media asset, and the agent handles the process of transcribing it into text. **How does `list_exports` help me with my videos?** `list_exports` lets you see a history of your renders. It tells you if an export is pending or complete, and provides the necessary link to download the finished media. **What's the difference between `get_project` and `get_transcription`?** `get_project` gives you metadata about the container (the project itself). `get_transcription` only pulls out the finished text from a specific media file within that project. **Can I initiate an export using `create_export`?** Yes, that's exactly what it does. You tell your agent which project to use and what format you want, and the tool kicks off the rendering process for you. **Do I need to use `list_drives` first before running other commands?** Yes, listing drives sets your context for Descript. It tells your agent which team drive containers you have access to. This scope is necessary so the system knows where to look when you ask it to run tools like get_project. **How does `list_templates` help me with my content planning?** It shows all available Descript project templates built into your account. This means your agent can pull up pre-set starting points for new projects, which speeds up setup time if you're building similar videos. **What happens if I use `get_project` and it fails?** The system will return a detailed error message. This tells you exactly why the request failed—for instance, if the project ID is incorrect or if your connection lacks the required permissions to view that specific file.