# Muse4AI

> Muse4AI is a free Windows app that lets AI agents (Claude Code, Codex, Cursor and others) create images, short videos and answers with the user's own Muse account, through MCP (http://127.0.0.1:47821/mcp) or REST (http://127.0.0.1:47821/v1). Independent project by Andrew Babo; not affiliated with Meta or Muse. Generation uses the user's Muse usage limit.

Agents: check status with `muse_status` first, give every new request a unique `request_id`, wait with `muse_get_job`, and never resend a job that is running or needs review.

## Agent skill

- [Muse4AI Agent Skill (.zip)](https://muse4ai.lovable.app/downloads/muse4ai-skill.zip): SKILL.md with setup, tools, workflows and troubleshooting

## Start here

- [What is Muse4AI?](https://muse4ai.lovable.app/docs/what-is-muse4ai.md): Learn how Muse4AI connects your Windows PC, your Muse account, and AI agents, and what using it means for your usage limit.
- [Install Muse4AI](https://muse4ai.lovable.app/docs/install.md): Install Muse4AI with the Windows EXE or ZIP package, handle SmartScreen safely, and keep your data when updating.
- [First open](https://muse4ai.lovable.app/docs/first-open.md): Complete first-run setup, choose where results are saved, and prepare a Muse account before sending your first request.

## Add your Muse account

- [Add a Muse account](https://muse4ai.lovable.app/docs/add-muse-account.md): Add a Muse account with its own proxy, sign in yourself, and check that the account is ready to accept jobs.

## Create in Studio

- [Create an image](https://muse4ai.lovable.app/docs/create-image.md): Create an image from a clear description in Studio, follow its progress, and find the finished file on your PC.
- [Create a video](https://muse4ai.lovable.app/docs/create-video.md): Describe a short video in Studio, submit it to Muse, and review the finished clip before using it in your project.
- [Ask and research](https://muse4ai.lovable.app/docs/ask-and-research.md): Ask Muse a question or attach research files in Studio, then read the answer and continue the same conversation.
- [Use reference files](https://muse4ai.lovable.app/docs/reference-files.md): Attach reference images, video, or research files in the right Studio mode and check them before submitting a job.

## Queue

- [Follow jobs in Queue](https://muse4ai.lovable.app/docs/queue.md): Track waiting and running jobs, preview finished results, and handle requests that need review without creating duplicates.

## Batches

- [Create and follow batches](https://muse4ai.lovable.app/docs/batches.md): Turn a pasted or imported prompt list into a batch, check each row, and follow progress without sending jobs one by one.

## Library

- [Find and reuse results](https://muse4ai.lovable.app/docs/library.md): Find finished results in Library, filter by prompt or batch, open local files, and reuse an image as a Studio reference.

## Accounts & usage limits

- [Accounts and usage limits](https://muse4ai.lovable.app/docs/accounts-and-limits.md): Check each Muse account’s usage, paused state, and parallel jobs before sending more work from the app or an agent.
- [Proxies and connection modes](https://muse4ai.lovable.app/docs/proxies-and-modes.md): Choose a proxy and connection mode for each account, and understand when Auto pauses instead of switching paths.

## Connect AI agents

- [Connect your AI agent](https://muse4ai.lovable.app/docs/connect-overview.md): Connect an agent on this PC, a cloud agent, or an advanced command bridge with the right address and a separate key.
- [Create and replace API keys](https://muse4ai.lovable.app/docs/api-keys.md): Create one key per agent, copy it while it is visible, and replace or revoke it safely if access needs to change.
- [Teach your agent with the Muse4AI skill](https://muse4ai.lovable.app/docs/agent-skill.md): Give your agent the Muse4AI skill or a short instruction note so it sends requests the safe way, without duplicates.
- [MCP tool reference](https://muse4ai.lovable.app/docs/mcp-tools.md): Learn the MCP tools, follow a job through its states, and avoid duplicate submissions when a request is interrupted.
- [REST API reference](https://muse4ai.lovable.app/docs/rest-api.md): Use the local REST API with a Bearer key, inspect its OpenAPI document, and follow jobs without submitting duplicates.

## Settings & remote access

- [Remote access](https://muse4ai.lovable.app/docs/remote-access.md): Let a cloud agent reach your PC through a tunnel or HTTPS domain, or share only within a trusted private network.
- [Find the right setting](https://muse4ai.lovable.app/docs/settings-overview.md): Find language, output folders, job controls, notifications, agent access, security, and diagnostics in Settings.
- [Updates, rollback, and backups](https://muse4ai.lovable.app/docs/updates-and-backups.md): Install a newer version without deleting your data, understand rollback limits, and recover from a checked backup.
- [Privacy and your data](https://muse4ai.lovable.app/docs/privacy-and-data.md): See what stays on your PC, what your prompts and attached files send to Muse, and how to limit agent access safely.

## Troubleshooting

- [Troubleshooting](https://muse4ai.lovable.app/docs/troubleshooting.md): Resolve common sign-in, queue, connection, and installation problems without accidentally sending the same job twice.
- [Glossary](https://muse4ai.lovable.app/docs/glossary.md): Look up unfamiliar app and agent terms in plain language, then return to the guide you were reading with more context.

## Open source & credits

- [Open source and acknowledgments](https://muse4ai.lovable.app/docs/open-source.md): See the projects behind Muse4AI, starting with muse2api, and find acknowledgments and installed license texts.

## Connect an agent

- [Claude Code](https://muse4ai.lovable.app/agents/claude-code): Connect Claude Code on your Windows PC to Muse4AI with one terminal command, then ask it to check your Muse tools.
- [Claude (desktop, web & mobile)](https://muse4ai.lovable.app/agents/claude): Give Claude on desktop, web, or mobile access to Muse4AI through a public HTTPS connector with your own API key.
- [Cursor](https://muse4ai.lovable.app/agents/cursor): Add Muse4AI to Cursor's MCP configuration so your coding agent can check Muse status and request media from your PC.
- [Codex](https://muse4ai.lovable.app/agents/codex): Connect a local Codex session to Muse4AI through its TOML settings and keep your Muse instructions in AGENTS.md.
- [GitHub Copilot in VS Code](https://muse4ai.lovable.app/agents/vscode-copilot): Add Muse4AI to your VS Code workspace so GitHub Copilot can call Muse tools while helping you work on a project.
- [Google Antigravity](https://muse4ai.lovable.app/agents/antigravity): Connect Google Antigravity to Muse4AI using its serverUrl configuration, then make a small status check before creating media.
- [Lovable](https://muse4ai.lovable.app/agents/lovable): Let the Lovable building agent use Muse4AI through a public connector when you need images or other Muse outputs for a project.
- [Hermes Agent](https://muse4ai.lovable.app/agents/hermes-agent): Add Muse4AI to Hermes Agent's YAML settings so your local assistant can inspect Muse status and use the available tools.
- [OpenClaw](https://muse4ai.lovable.app/agents/openclaw): Connect OpenClaw's Gateway to Muse4AI using an address the Gateway can reach, whether it runs on your PC or elsewhere.
- [OpenCode](https://muse4ai.lovable.app/agents/opencode): Configure Muse4AI as a remote-type MCP server in OpenCode while keeping requests on the same PC as your Windows app.
- [LM Studio](https://muse4ai.lovable.app/agents/lm-studio): Use Muse4AI from LM Studio by adding an MCP server and choosing a local model that can reliably call external tools.
- [n8n](https://muse4ai.lovable.app/agents/n8n): Connect an n8n AI Agent to Muse4AI with the MCP Client Tool node, choosing an endpoint reachable from your n8n instance.
- [Cline](https://muse4ai.lovable.app/agents/cline): Add Muse4AI to Cline's MCP settings and let your coding assistant check the available Muse tools before requesting outputs.
- [Warp](https://muse4ai.lovable.app/agents/warp): Add Muse4AI in Warp's agent settings so a local terminal agent can check Muse status and use your connected account's tools.
- [Gemini CLI](https://muse4ai.lovable.app/agents/gemini-cli): Configure Gemini CLI's HTTP MCP connection to Muse4AI, and check your account's CLI eligibility before following this setup.
- [Grok CLI](https://muse4ai.lovable.app/agents/grok-cli): Connect Grok CLI to Muse4AI with an HTTP MCP command and your API key; this guide does not cover the Grok chat app.
- [Manus](https://muse4ai.lovable.app/agents/manus): Give Manus access to Muse4AI through a custom MCP server with a public HTTPS address and a separate Bearer API key.
- [Windsurf (Cascade)](https://muse4ai.lovable.app/agents/windsurf): Add Muse4AI to Cascade's MCP configuration with serverUrl and an Authorization header, then refresh the available tools.
- [ChatGPT](https://muse4ai.lovable.app/agents/chatgpt): ChatGPT cannot yet accept Muse4AI's Bearer key in its custom connector flow, so this connection is listed for later.

## Use cases

- [Affiliate videos with clear product details](https://muse4ai.lovable.app/use-cases/affiliate-videos): Plan affiliate videos from real product photos, then generate supporting clips for TikTok Shop edits without inventing product claims.
- [Product photos for a consistent shop set](https://muse4ai.lovable.app/use-cases/product-photos): Create AI product photo concepts from real references, then check shape, color, and packaging before using them in your shop.
- [Lesson videos with reviewed illustrations](https://muse4ai.lovable.app/use-cases/lesson-videos): Prepare lesson videos with AI illustrations and short opening clips, while keeping explanations and classroom examples under teacher review.
- [Short ads built from approved scenes](https://muse4ai.lovable.app/use-cases/short-ads): Create short AI ad concepts with separate hooks, product scenes, and closing shots, then assemble approved clips in your editor.
- [Sales videos with offers you can verify](https://muse4ai.lovable.app/use-cases/sales-videos): Build sales videos from your current prices, offer terms, and real product references, with AI clips reserved for supporting scenes.
- [B-roll for agent-built Remotion videos](https://muse4ai.lovable.app/use-cases/remotion-broll): Have your agent build a Remotion composition and request B-roll clips or images through Muse4AI, keeping timing and captions editable.
- [Storyboards to review before production](https://muse4ai.lovable.app/use-cases/storyboards): Use an AI storyboard workflow to turn a script into shot references and rough motion tests before committing to production.
- [Report videos with charts from your data](https://muse4ai.lovable.app/use-cases/data-report-videos): Turn a sales spreadsheet into a data-report video plan, with charts made in a chart tool and AI visuals used only as supporting scenes.
- [Realistic B-roll, clearly illustrative](https://muse4ai.lovable.app/use-cases/realistic-broll): Generate realistic B-roll for narration gaps and abstract topics, with clear labels so illustrative scenes are not mistaken for evidence.
- [Menu visuals grounded in real dishes](https://muse4ai.lovable.app/use-cases/restaurant-menus): Explore AI menu visuals from photos of your real dishes, then check portions and ingredients before adding prices in a layout tool.
- [Social campaign visuals from one brief](https://muse4ai.lovable.app/use-cases/social-campaigns): Plan a social campaign with related AI image concepts and supporting clips, then adapt each approved asset to its posting format.
- [Thumbnail backgrounds with room for text](https://muse4ai.lovable.app/use-cases/thumbnails): Create AI thumbnail background options around your video's actual topic, then add readable titles and brand elements in an editor.
- [Onboarding visuals for real procedures](https://muse4ai.lovable.app/use-cases/hr-onboarding): Turn approved HR procedures into an onboarding video plan, using AI illustrations for generic scenes and real screenshots for exact steps.
- [Property clips with honest visual context](https://muse4ai.lovable.app/use-cases/real-estate-clips): Plan real-estate clips around verified listing details and real property footage, using clearly labeled AI scenes only for illustration.
- [A Shorts series planned before batching](https://muse4ai.lovable.app/use-cases/shorts-series): Plan a YouTube Shorts series with repeatable scripts and batched AI scene requests, then review each episode before editing and publishing.
- [NGO explainers with responsible visuals](https://muse4ai.lovable.app/use-cases/ngo-explainers): Build NGO explainer videos from approved source material, using labeled AI illustrations rather than fabricated beneficiary stories or footage.
- [Event recaps built around real footage](https://muse4ai.lovable.app/use-cases/event-recaps): Plan event recap videos around real recordings, with AI-generated opening and closing visuals kept separate from evidence of the event.
- [Research notes into a visual brief](https://muse4ai.lovable.app/use-cases/research-to-visuals): Turn research notes into a visual brief with source-linked claims, illustrative image ideas, and a review step before generating assets.

## Optional

- [All guides in one file](https://muse4ai.lovable.app/llms-full.txt)
- [Tiếng Việt](https://muse4ai.lovable.app/vi/docs)