PluginPartly tested (needs an account)
fal.ai creative media for Gemini
fal-gemini-extension
Gemini CLI extension that connects fal's hosted MCP server to make images, video, audio and 3D, with six creative workflow skills. Runs bill fal credits.
- Works with
- Gemini CLI
- Good for
- Graphic designers, Build an app
What it does
This extension connects Gemini CLI to fal's hosted MCP server. fal runs many image, video, audio and 3D models. Gemini can then pick a model, check its price, send the job and fetch the result for you.
You sign in to your fal account in the browser with /mcp auth fal. You do not put an API key in a file. The MCP server is free, but each model run is billed to your fal account at fal's normal prices.
The extension adds six skills. fal-media handles model choice, prices, uploads and job recovery. Five guides cover product ad campaigns, game assets, storyboard to short film, product photography, and localization. A larger skill library is in the separate fal-ai-community/skills repository.
Example prompts
- "Use fal to plan a three-image campaign for my reusable water bottle. Ask for the product reference and missing brand details. Check models and prices, then show the deliverables and estimated cost before generation." (the README example)
The README also lists these jobs: a product campaign, game sprites with an atlas, a short film from a storyboard, product photography from real product photos, and localization with captions.
How to install
gemini extensions install https://github.com/fal-ai-community/fal-gemini-extensionHow to use it
Confirm the install prompt, restart Gemini CLI, run /mcp auth fal and sign in to fal in the browser. Then ask for a creative task. Each model run bills your fal account.
Check it works
- Run
gemini extensions listand look forfal-ai. - Restart Gemini CLI and run
/mcp auth fal. Sign in to fal in the browser. - Run
/mcpand check that the fal tools appear before you ask for any media.
Common problems
- Gemini CLI does not let you sign in — since 18 June 2026 Gemini CLI does not work with personal Google accounts. It works with a company Gemini Code Assist Standard or Enterprise licence, a paid Gemini API key or Vertex AI. See how to install Gemini CLI.
- fal tools are missing — restart Gemini CLI, run
/mcp auth falagain, and update Gemini CLI. The README asks for a current version with remote Streamable HTTP and browser OAuth — https://github.com/fal-ai-community/fal-gemini-extension - The sign-in page does not open — OAuth needs a browser on the same machine. It does not work on a headless server or some SSH sessions — https://geminicli.com/docs/tools/mcp-server/
- Gemini uses a different fal setup — a server named
falin your Gemini settings wins over the extension. Review or remove that entry — https://geminicli.com/docs/extensions/reference/ - You need every fal skill — use the community repository's install guide — https://github.com/fal-ai-community/skills/blob/main/docs/install.md
What to watch out for
- Since 18 June 2026 Gemini CLI does not work with personal Google accounts; it works with a company Gemini Code Assist Standard or Enterprise licence, a paid Gemini API key or Vertex AI
- Needs a fal account; the MCP server is free, but every model run your assistant triggers is billed at standard fal pricing
- Sends your prompts and any files you ask it to use (uploaded as source images or audio) to fal's remote MCP server at https://mcp.fal.ai/mcp-relay
- Sign-in is browser OAuth and needs a local browser; it does not work on a headless machine or some SSH sessions
- Loads GEMINI.md into every Gemini CLI session and adds six skills that Gemini can pick on its own when you ask for media work
- If your Gemini settings already define an MCP server named fal, your settings win over the extension
- No CLI, renderer or hooks are installed; finished films, atlases and caption exports may need local tools you do not have
- Repository created 2026-11-07 with no other users yet; the author says OAuth sign-in and generation were not yet tested end to end
Licence and author
- Licence
- MIT
- Author
- fal-ai-community
How we tested it
We installed it on an empty test computer on 7 Nov 2026. Partly tested: installed in a clean test system, but its main features need an account or key we did not use.
Not installed or run. We have no paid Gemini CLI account or fal credits for a test.
Technical details of the test
- Date
- 7 Nov 2026
- Method
- Reviewed by our team against the repository at 8dee698, its README, GEMINI.md and the six SKILL.md files, the Gemini CLI extension and MCP docs, and fal's MCP page.
Checked again on 7 Nov 2026
Alternatives
- Nano Banana — Gemini CLI extension for Google's own image models, paid with a Gemini API key
- HyperFrames — builds short videos as web pages and renders them to MP4; works with Gemini CLI too
FAQ
- Q: Do I need an API key? A: No. The README says you sign in with
/mcp auth fal. fal gives a key only to clients that cannot use MCP sign-in. - Q: Is it free? A: The extension and fal's MCP server are free. fal bills each model run. Ask Gemini to show the price before it generates.
- Q: Does it work with my personal Google account? A: No. Since 18 June 2026 Gemini CLI does not work with personal Google accounts.
- Q: Is this Antigravity? A: No. It is a Gemini CLI extension.
- Q: How do I update or remove it? A: Run
gemini extensions update fal-ai(README) orgemini extensions uninstall fal-ai.
Sources
- github.com/fal-ai-community/fal-gemini-extension
- github.com/fal-ai-community/fal-gemini-extension/blob/8dee6982494260a16dc381ce5d4648b0522f91c6/README.md
- github.com/fal-ai-community/fal-gemini-extension/blob/8dee6982494260a16dc381ce5d4648b0522f91c6/gemini-extension.json
- github.com/fal-ai-community/fal-gemini-extension/blob/8dee6982494260a16dc381ce5d4648b0522f91c6/GEMINI.md
- github.com/fal-ai-community/fal-gemini-extension/blob/8dee6982494260a16dc381ce5d4648b0522f91c6/skills/fal-media/SKILL.md
- fal.ai/mcp
- geminicli.com/docs/extensions/reference/
- geminicli.com/docs/tools/mcp-server/