Assistants
Use Audo in ChatGPT, Claude and more.
Add it once. Then ask for audio work in plain words.
In ChatGPT on the web, open Settings, then Security and login, and turn on Developer mode. It's for Plus, Pro, Business, Enterprise, and Edu.
Open chatgpt.com/plugins, click +, and create an app with this address:
https://audo.ai/mcpSign in to Audo when ChatGPT asks.
In a chat, choose Developer mode from the + menu and pick Audo. Then attach a file and ask.
Your files
Attach a file to the chat and ChatGPT hands it to Audo. For large files, or in the mobile app, ChatGPT gives you an upload link instead.
Click Add to Claude, then Add. It opens Claude's custom connector form with Audo filled in.
Or add it by hand: open Customize, then Connectors, click +, choose Add custom connector, and paste this address:
https://audo.ai/mcpSign in to Audo when Claude asks.
In a chat, turn on Audo from the + menu under Connectors, then ask.
Your files
An upload box appears in the chat. Works on the web, desktop, and mobile, on every plan (Free plans can add one custom connector). On Team and Enterprise, an owner adds Audo in Organization settings, then members connect it.
Add Audo to Claude. Cowork uses the connectors on your Claude account, so there's nothing else to set up.
Or add it by hand: open Customize, then Connectors, click +, choose Add custom connector, and paste this address:
https://audo.ai/mcpSign in to Audo when Claude asks, then give Cowork the task.
Your files
Cowork uploads your files with curl when network egress allows audo.ai. Otherwise it shows an upload box, or gives you an upload link.
Allow audo.ai once
So Cowork can upload and download files without asking each time. To let Cowork upload and download files itself, turn on network egress in Settings, then Capabilities. On Team and Enterprise, an owner adds audo.ai to the allowed domains instead.
Add Audo from your terminal:
Terminalclaude mcp add --transport http --scope user audo https://audo.ai/mcpSign in. Run /mcp in Claude Code, pick audo, and approve in your browser.
Your files
Claude Code uploads your files to Audo with curl, and saves each result next to the original.
Allow audo.ai once
So Claude Code can upload and download files without asking each time. This turns on Claude Code's sandbox: commands run without asking, and they can reach audo.ai. Merge it with what's already in the file.
{
"sandbox": {
"enabled": true,
"network": {
"allowedDomains": [
"audo.ai"
]
}
}
}Add Audo:
Terminalcodex mcp add audo --url https://audo.ai/mcpSign in:
Terminalcodex mcp login audo
Your files
Codex uploads your files to Audo with curl, and saves each result next to the original.
Allow audo.ai once
So Codex can upload and download files without asking each time. This lets Codex run curl without asking, outside its sandbox, for any address. Restart Codex after saving it.
prefix_rule(pattern = ["curl"], decision = "allow", justification = "Upload and download files at audo.ai")Click Add to Cursor, then Install. Or add Audo to ~/.cursor/mcp.json yourself:
~/.cursor/mcp.json{ "mcpServers": { "audo": { "url": "https://audo.ai/mcp" } } }Sign in. Cursor shows a Connect button next to Audo in its MCP settings the first time.
Your files
Cursor uploads your files to Audo with curl, and saves each result next to the original.
Allow audo.ai once
So Cursor can upload and download files without asking each time. This lets commands in Cursor's sandbox reach audo.ai. Merge it with what's already in the file.
{
"networkPolicy": {
"allow": [
"*.audo.ai"
]
}
}Add Audo:
Terminalgemini mcp add --transport http --scope user audo https://audo.ai/mcpSign in. Run /mcp auth audo in Gemini CLI and approve in your browser.
Your files
Gemini CLI uploads your files to Audo with curl, and saves each result next to the original.
Allow audo.ai once
So Gemini CLI can upload and download files without asking each time. This lets Gemini CLI run curl without asking, for any address.
[[rule]]
toolName = "run_shell_command"
commandPrefix = "curl"
decision = "allow"
priority = 100Click Add to VS Code, then Install. Or run this:
Terminalcode --add-mcp '{"name":"audo","type":"http","url":"https://audo.ai/mcp"}'Sign in. VS Code opens your browser the first time it connects.
Your files
Agent mode uploads your files to Audo with curl, and saves each result next to the original.
Allow audo.ai once
So VS Code can upload and download files without asking each time. This lets agent mode run curl without asking, for any address.
{
"chat.tools.terminal.autoApprove": {
"curl": true
}
}Add Audo:
Terminalcopilot mcp add --transport http audo https://audo.ai/mcpSign in. Run /mcp auth in Copilot CLI and approve in your browser.
Your files
Copilot CLI uploads your files to Audo with curl, and saves each result next to the original.
Allow audo.ai once
So Copilot CLI can upload and download files without asking each time. Starting it this way lets Copilot run curl without asking, for any address.
copilot --allow-tool='shell(curl:*)'Add Audo to ~/.gemini/config/mcp_config.json: Open it from the agent panel's … menu: MCP Servers, then Manage MCP Servers, then View raw config.
mcp_config.json{ "mcpServers": { "audo": { "serverUrl": "https://audo.ai/mcp" } } }Sign in. In Settings, open Customizations and click Authenticate next to audo. Approve in your browser, paste the code it shows back into Antigravity, and click Submit.
Your files
Antigravity uploads your files to Audo with curl, and saves each result next to the original.
Add Audo to Zed's settings.json: Open it with the command zed: open settings file. Or open Settings, then AI, then MCP Servers, and choose Add Remote Server.
settings.json{ "context_servers": { "audo": { "url": "https://audo.ai/mcp" } } }Sign in. Zed asks you to sign in to Audo when it connects. A green dot means it's ready.
Your files
Zed's agent uploads your files to Audo with curl, and saves each result next to the original.
Allow audo.ai once
So Zed can upload and download files without asking each time. This lets the agent run curl commands for audo.ai without asking. Merge it with what's already in the file.
{
"agent": {
"tool_permissions": {
"tools": {
"terminal": {
"always_allow": [
{
"pattern": "^curl\\s+.*audo\\.ai"
}
]
}
}
}
}
}Add Audo to ~/.config/opencode/opencode.json:
opencode.json{ "$schema": "https://opencode.ai/config.json", "mcp": { "audo": { "type": "remote", "url": "https://audo.ai/mcp", "enabled": true } } }Sign in:
Terminalopencode mcp auth audo
Your files
OpenCode uploads your files to Audo with curl, and saves each result next to the original.
Allow audo.ai once
So OpenCode can upload and download files without asking each time. This lets OpenCode run curl commands for audo.ai without asking. Merge it with what's already in the file.
{
"permission": {
"bash": {
"curl *audo.ai*": "allow"
}
}
}In Raycast, run the Install MCP Server command.
Name it Audo, choose HTTP, paste this address, set OAuth to Dynamic, and press Install MCP Server:
https://audo.ai/mcpChoose Sign In, and approve in your browser.
Your files
Raycast gives you an upload link for each file, and a download link for the result.
Then just ask.
Your assistant gets the same free use. For longer files, see pricing.
Remove the background noise from interview.wav, then transcribe it.
Remove the long pauses from episode-12.mp3 and make it podcast-loud.
Make SRT captions for launch-video.mp4.
Pull the audio out of this video as an MP3.
Building your own agent? Read the docs.