Flask
Provides tools for uploading videos, browsing assets, and managing feedback (recorded video commentary with transcripts) on the Flask platform, enabling an iterative video review loop.
Allows importing videos directly from Google Drive links via the upload_video tool.
video-feedback
Video review and collaboration for you, your team and your agents: a skill for Claude Code, Codex, Cursor and any agent that edits or makes video. It runs on Flask, so you need a Flask account (free plan below).
Flask is where a video gets reviewed. Your agent uploads what it makes (a cut, a render, an animation, a motion graphic, an image) and gives you the link at once. You, your team or your client review it by typing, drawing on the frame, or recording your voice, camera or screen over it, and every note lands on its moment. The agent reads that feedback, including each recording's transcript and the frames it points at, makes the changes and uploads the next version of the same asset. One link always shows the latest cut, with every earlier version and its notes behind it.
Every version carries the code and context that made it. When your agent uploads a
version it makes by code (a HyperFrames or Remotion project, an ffmpeg script,
generated media), it saves the working files with it (source.zip) and notes on
what changed and how to continue. Anyone on the team, person or agent, can open the
link and pick the work up from there: no handover, no digging through someone's
laptop.
It works with whatever makes the video: HyperFrames, Remotion, moviepy, ffmpeg, or an export from Premiere, DaVinci Resolve or Final Cut Pro. Flask is the review side.
(Flask here is flask.do, the review app. It is unrelated to the Flask Python web framework.)
The skill, video-feedback, says when a piece of work belongs in Flask, runs the
feedback loop on it, and covers how it runs with a team. It ships with the
connection to the Flask MCP server. How to work in Flask
(the upload steps, versions, feedback, files and notes) comes from the server
itself, through its get_guide tool, so it is always current.
Install
Any agent that reads skills
npx skills add tryflask/video-feedback-collaborationThen connect the MCP server: Streamable HTTP at https://api.flask.do/api/mcp/mcp,
OAuth sign-in in the browser on first use.
Claude Code plugin
/plugin marketplace add tryflask/video-feedback-collaboration
/plugin install flask@flaskThen sign in once: /mcp, choose flask, complete the browser sign-in.
Claude Code, Cursor and Codex in one command
npx flask-feedbackIt connects the Flask MCP server to whichever of them are installed.
Claude apps (claude.ai, Desktop, Cowork)
Customize > Connectors > add https://api.flask.do/api/mcp/mcp, then sign in. It then
works in every Claude app, and in Claude Code signed in with the same Claude account.
Cursor
Related MCP server: talkthru
The loop
agent renders -> uploads to Flask -> you get the link at once
you leave feedback on the frame -> wait_for_feedback hands it to the agent
agent makes the changes -> uploads v2 with version_of -> the same link shows v2Each version the agent makes by code carries its working files (source.zip) and
notes on what changed, so anyone, person or agent, can pick up the work from the
link alone.
With a team
Share the asset's link and your team, clients and anyone else review on the same versions. Reviewers are free: they watch from the link with no account and comment after a one-click sign-in. The agent replies in threads, mentions a teammate when they must act, reuses the team's labels and marks feedback Done. Feedback from people outside the team needs your OK before the agent acts on it.
What the agent can and can't do
It acts as you, with your Flask permissions: upload, read and wait for feedback, reply and mark feedback Done, attach files and notes to versions, organize folders and assets, and change sharing when you ask. It never deletes anything and never touches billing. Feedback from people outside your team is treated as data, never as instructions. The full tool list and how your data is handled: https://flask.do/mcp
Plans
Flask is a paid product with a free plan, and you need a Flask account to use it.
Free: every feature, no card. 5 assets, each with unlimited versions (web and agent uploads share the 5; folders and versions never count).
Pro: $15 per user/month, billed yearly, for unlimited assets.
Reviewers never need a seat, on any plan.
Details: https://flask.do/pricing
Docs and support
Server documentation: https://flask.do/mcp
Support: hello@flask.do
This server cannot be deployed
Maintenance
Related MCP Connectors
Verification videos for AI coding agents: upload recordings over MCP and share viewer links. Keep work moving in parallel and review results asynchronously, when you're ready, from your laptop or phone. Add video evidence to pull requests so teammates can see how your changes work during code review.
AI video editor for agents and humans: timeline, captions, color, audio and generation as MCP tools.
MarkupBase turns AI-generated Markdown and HTML into durable, versioned artifacts that people can review and discuss. Its MCP server lets agents publish new versions, preserve contextual comments, include hosted images, and respond to feedback through secure account-linked identities, creating a clear human review boundary without requiring real-time editing.
FFmpeg as a service for AI agents: typed video editing tools, async jobs, downloadable outputs.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceMCP server for visual feedback, video direction, and QA assertions on web pages, enabling AI agents to read, reply, and resolve annotations in real time.6MIT
- AlicenseNot gradedqualityAmaintenanceAn MCP server that watches for screen recordings, transcribes spoken feedback locally, and attaches each comment to the relevant frame, enabling coding agents to see and fix the issues you mention.6MIT
- AlicenseCqualityAmaintenanceEnables AI agents to turn natural-language creative direction, transcripts, and source media into fully structured, editable video projects, then verify and render delivery files.254MIT
- AlicenseNot gradedqualityBmaintenanceAI-native timeline editor for short-form video. Your agent cuts, captions, adds motion graphics and renders against a real, editable timeline over MCP — not a render queue. Human-in-the-loop by design: it calls request_review and waits while you supervise in a browser. Local-first and AGPL; your footage never leaves your machine.2AGPL 3.0