Run your streams by asking

Stream Repeater has an MCP server, so Claude, ChatGPT, Cursor, VS Code and other AI assistants can check your stream, go live, pause a destination, clip a moment or read chat when you ask them to. Sign in from Claude or ChatGPT, or connect any other client with your own API token; either way the assistant can do nothing you could not.

Included on every plan. Claude and ChatGPT sign in with OAuth; any other MCP client uses an API token.

Your AI assistantStream Repeater connected
  1. You: Is my stream healthy right now?
  2. Used get_channel_status · The Signal
  3. Assistant: Yes. The Signal has been live for 1 h 12 min at about 6 Mb/s with no packet loss. YouTube and Twitch are running; Kick is starting again after a drop. Want me to check it in a minute?
An example conversation. The tools are real; your assistant will word things its own way.

What it is, in plain language

MCP, the Model Context Protocol, is how AI assistants use other software. Connect ours once and your assistant can see and run your Stream Repeater workspace the same way you do in the dashboard.

  1. You ask

    Type what you want, the way you would ask a producer. No commands to learn.

  2. It uses your API

    The assistant picks the right tools, and each one calls the same API a script would, with the same permission checks.

  3. You stay in charge

    The server tells assistants to check with you before anything viewers would notice, and a read-only token cannot change anything at all.

Things you can ask

A few conversations, with the tools the assistant used along the way.

Pause one platform, keep the rest

Your AI assistantStream Repeater connected
  1. You: Pause YouTube for now but keep Twitch running.
  2. Used list_destinations · The Signal
  3. Assistant: Pausing YouTube ends the broadcast for viewers there. Go ahead?
  4. You: Yes.
  5. Used set_destination_enabled · YouTube → paused
  6. Assistant: YouTube is paused. Twitch and Kick are still live. Tell me when to bring it back.

Book a Podcast Room

Your AI assistantStream Repeater connected
  1. You: Book a podcast room on The Signal for Friday at 8pm and give me the link to send.
  2. Used start_podcast_room · Fri 20:00
  3. Assistant: Booked for Friday at 20:00. The code is K7M2QX and the link is streamrepeater.com/pod/K7M2QX. Guests wait until you let them in, and you go live from the studio.
  4. You: It's Friday. Is anyone waiting?
  5. Used get_podcast_room · K7M2QX
  6. Assistant: Priya Nair is in the waiting room. Shall I let her in?

Clip a moment

Your AI assistantStream Repeater connected
  1. You: Clip the last 30 seconds and call it "Clutch round".
  2. Used create_clip · 30 s
  3. Assistant: Done. "Clutch round" is being cut from the live recording and will be in your clips in a moment.

Your keys stay yours

Your AI assistantStream Repeater connected
  1. You: What is my stream key? I need it for OBS.
  2. Used get_stream_keys · The Signal
  3. Assistant: I can only see a masked key, so I cannot read it out. Copy it from the channel page in the dashboard, under the connection details. The server address for OBS is fine to use as it is.

An example conversation. The tools are real; your assistant will word things its own way.

Everything your assistant can do

The server has 35 tools today. This list is read from the server itself, so it is always the current one.

Workspaces and health

5

What you have, whether it is healthy, and how much you have used.

  • list_workspaces Read-only

    Lists your workspaces, your role in each and its plan.

  • list_channels Read-only

    Lists channels with live status, the go-live switch, recording and destination counts.

  • get_channel_status Read-only

    Health of a stream: live or not, bitrate, uptime, packet loss, viewers and every destination's relay state.

  • get_usage Read-only

    Bandwidth used against your allowance, with a forecast.

  • get_stream_keys Read-only

    Server addresses for your encoder, with keys masked, and where to copy the real ones.

Going live and destinations

6

Switch a channel on or off, add, pause and resume platforms.

  • list_destinations Read-only

    Lists where a channel is relayed to, without URLs or stream keys.

  • get_live_station_meta Read-only

    Retrieve public metadata for a published AOD recording: title, genre, stream URL, duration. Uses the recording's public_token (the last segment of its listen URL). No auth required.

  • go_live Makes changes

    Switches a channel's ingest on, so it relays to every enabled destination.

  • end_stream Makes changes

    Switches a channel's ingest off, ending the broadcast everywhere.

  • add_destination Makes changes

    Adds a destination from a connected account, a full URL, or a server and key.

  • set_destination_enabled Makes changes

    Pauses or resumes one destination.

Recordings and clips

8

Turn cloud recording on or off, list recordings and make clips.

  • list_recordings Read-only

    Lists a channel's cloud-recorded segments.

  • list_audio_recordings Read-only

    List the Audio on Demand (AOD) recordings for an icecast_ingest channel. Returns each recording's title, duration, public access state and listen URL. Only available on audio-workspace channels.

  • get_recording_stats Read-only

    Get play analytics for a channel's Audio on Demand recordings: total plays, plays by day, and top listener countries over a time window (default 30 days, max 90).

  • list_clips Read-only

    Lists a channel's clips, newest first.

  • start_recording Makes changes

    Turns cloud recording on for a channel.

  • stop_recording Makes changes

    Turns cloud recording off for a channel.

  • create_clip Makes changes

    Clips the last few seconds of the live recording, or cuts a clip from a recording.

  • toggle_recording_public Makes changes

    Make an Audio on Demand recording publicly accessible (or private). When public, anyone with the link can listen. Confirm with the user before making a recording public.

Live chat

2

Read your unified chat and post in it.

  • get_recent_chat Read-only

    Reads recent messages from your unified live chat.

  • send_chat_message Makes changes

    Posts a message in your live chat on the connected platforms.

Podcast Rooms

12

Open or book rooms, see who is waiting, admit, mute, stage or remove guests. Podcast Rooms →

  • list_podcast_rooms Read-only

    Lists a channel's Podcast Rooms: open, booked or past, with join codes and links.

  • get_podcast_room Read-only

    One Podcast Room: status, join link, and who is waiting or on air.

  • list_podcast_participants Read-only

    Lists the host and guests in a Podcast Room, waiting or on air.

  • get_podcast_settings Read-only

    A channel's Podcast Rooms defaults, branding and custom domain state.

  • start_podcast_room Makes changes

    Opens a Podcast Room now, or books one for later, and returns the join link.

  • end_podcast_room Makes changes

    Ends an open Podcast Room for everyone, or cancels a booked one.

  • admit_podcast_guest Makes changes

    Admits a guest from the waiting room.

  • deny_podcast_guest Makes changes

    Turns away a guest who is still waiting.

  • remove_podcast_guest Makes changes

    Takes a guest off air and out of the room.

  • mute_podcast_guest Makes changes

    Mutes or unmutes a guest in the host's mix.

  • stage_podcast_guest Makes changes

    Brings a guest on stage or sends them backstage.

  • update_podcast_settings Makes changes

    Changes a channel's Podcast Rooms defaults.

More

2

Other tools on the server.

  • get_station_status Read-only

    Get the current listener count, today's peak, broadcast bitrate, and what is currently on air for an Icecast ingest channel. Returns 404 for non-radio channels. Data comes from the most recent 5-minute snapshot.

  • get_station_track_history Read-only

    Get the last 50 tracks played on a radio station (Icecast ingest channel), most recent first. Each entry includes title, artist, and when it started playing. Returns 404 for non-radio channels.

Full tool reference and client setup →

Safe to hand to an assistant

The assistant works through an API token you create and can revoke. It gets no more access than that token, and less than you have.

  • Read-only or full accessChoose when you create the token. A read-only token can check health, usage and chat but cannot change anything, and your workspace role still applies.
  • Keys are never sharedStream keys, destination URLs and passwords are never returned to an assistant. Connection details come back masked.
  • Every change is loggedEach change made through MCP appears in your workspace Activity log, naming the tool, the token and the person.
  • Asks before it actsThe server tells assistants to confirm with you before ending a stream, pausing a platform, posting in chat or admitting a guest.
  • Revoke any timeRevoke the token in Settings and the assistant loses access straight away. Each token is limited to 120 MCP requests a minute.
  • Chat treated as dataViewers write chat messages, so the server tells assistants never to follow instructions found in them.
The workspace Activity log: "An AI assistant made a change. AI assistant (MCP): opened a podcast room on The Signal, token Claude Desktop, by Jordan Blake"
What a change made by an assistant looks like in your Activity log.

Set up with the assistant you use

Pick your client to see where the setting goes, the exact snippet and how to check it works. It takes about two minutes.

Claude Code

Anthropic's coding agent in your terminal. One command adds the server for every project.

Official setup guide (opens in a new tab) ↗
  1. Create a token

    In Settings → API tokens, create a token just for this assistant. Choose Read-only if it should only look.

    Sign in to create a token
  2. Add the server

    For you (all projects)
    ~/.claude.json
    For one project
    .mcp.json
    One command
    claude mcp add --transport http --scope user streamrepeater https://streamrepeater.com/mcp \
      --header "Authorization: Bearer YOUR_TOKEN"
    Config file, token read from an environment variable .mcp.json
    {
        "mcpServers": {
            "streamrepeater": {
                "type": "http",
                "url": "https://streamrepeater.com/mcp",
                "headers": {
                    "Authorization": "Bearer ${STREAMREPEATER_TOKEN}"
                }
            }
        }
    }
  3. Check it works

    Run this and look for a tick next to streamrepeater, or type /mcp inside a session:

    claude mcp list

    Then ask: List my channels and tell me which are live.

  • For a team project, commit .mcp.json with ${STREAMREPEATER_TOKEN} in it and have everyone set that variable themselves. Claude Code asks each person to approve a project server the first time.
  • Adding the server does not test the token. A wrong token shows up as a failed connection in claude mcp list.

Claude (web, desktop & mobile)

claude.ai, the Claude desktop app and Cowork share custom connectors, and the mobile apps use the ones you add there. Free plans can add one custom connector.

Official setup guide (opens in a new tab) ↗
  1. Add a custom connector

    1. Open Customize → Connectors, click + Add, then Add custom connector. On a Team or Enterprise plan, an owner adds it in Organization settings → Connectors instead.
    2. Enter Stream Repeater as the name and the server URL below, then click Continue.
    3. Under Authentication, keep the sign-in option Claude suggests, then click Add.
    4. Sign in to Stream Repeater when asked, choose a workspace, and choose whether the assistant can make changes.
    Server URL
    https://streamrepeater.com/mcp
  2. Check it works

    In a chat, click +, open Connectors and switch Stream Repeater on.

    Then ask: List my channels and tell me which are live.

  • Claude connects from Anthropic's servers, not from your computer, so it works the same on every device. You can see and disconnect it in Settings → Connected apps.
  • Prefer a token? Under Authentication, choose No sign in, then add Authorization: Bearer YOUR_TOKEN under Request headers. On Team and Enterprise plans, a header added by an owner is shared by everyone who uses the connector, so use a Read-only token unless everyone should be able to make changes.

OpenAI Codex

OpenAI's coding agent. The command line, the IDE extension and the Codex app share one config file.

Official setup guide (opens in a new tab) ↗
  1. Create a token

    In Settings → API tokens, create a token just for this assistant. Choose Read-only if it should only look.

    Sign in to create a token
  2. Add the server

    For you (all projects)
    ~/.codex/config.toml
    For one project
    .codex/config.toml
    Set the token as an environment variable first
    export STREAMREPEATER_TOKEN="YOUR_TOKEN"
    One command
    codex mcp add streamrepeater --url https://streamrepeater.com/mcp \
      --bearer-token-env-var STREAMREPEATER_TOKEN
    Config file, token read from an environment variable ~/.codex/config.toml
    [mcp_servers.streamrepeater]
    url = "https://streamrepeater.com/mcp"
    bearer_token_env_var = "STREAMREPEATER_TOKEN"
  3. Check it works

    Run this, or type /mcp inside Codex:

    codex mcp list

    Then ask: List my channels and tell me which are live.

  • Codex reads the token from the environment variable each time it starts, so the token never sits in the config file. Add the export line to your shell profile to keep it.
  • On Windows, set the variable with setx STREAMREPEATER_TOKEN "YOUR_TOKEN" and open a new terminal.

Gemini CLI

Google's open-source agent for the terminal.

Official setup guide (opens in a new tab) ↗
  1. Create a token

    In Settings → API tokens, create a token just for this assistant. Choose Read-only if it should only look.

    Sign in to create a token
  2. Add the server

    For you (all projects)
    ~/.gemini/settings.json
    For one project
    .gemini/settings.json
    One command
    gemini mcp add --transport http --scope user \
      --header "Authorization: Bearer YOUR_TOKEN" \
      streamrepeater https://streamrepeater.com/mcp
    Config file ~/.gemini/settings.json
    {
        "mcpServers": {
            "streamrepeater": {
                "httpUrl": "https://streamrepeater.com/mcp",
                "headers": {
                    "Authorization": "Bearer YOUR_TOKEN"
                }
            }
        }
    }
  3. Check it works

    Run this, or type /mcp inside a session:

    gemini mcp list

    Then ask: List my channels and tell me which are live.

  • Use httpUrl, not url: in Gemini CLI, url means the older SSE transport.
  • Without --scope user the server is added to the current project only.

Google Antigravity

Google's agent-first IDE and its agy command line.

Official setup guide (opens in a new tab) ↗
  1. Create a token

    In Settings → API tokens, create a token just for this assistant. Choose Read-only if it should only look.

    Sign in to create a token
  2. Add the server

    macOS and Linux
    ~/.gemini/config/mcp_config.json
    Windows
    %USERPROFILE%\.gemini\config\mcp_config.json
    Open it from
    … → MCP Servers → Manage MCP Servers → View raw config
    Config file mcp_config.json
    {
        "mcpServers": {
            "streamrepeater": {
                "url": "https://streamrepeater.com/mcp",
                "headers": {
                    "Authorization": "Bearer YOUR_TOKEN"
                }
            }
        }
    }
  3. Check it works

    In the Agent panel, open … → MCP Servers and refresh the list. Stream Repeater should appear with its tools.

    Then ask: List my channels and tell me which are live.

  • If your version keeps the file somewhere else, View raw config always opens the right one.

Cursor

The AI code editor. A global file works in every project; a project file only in that project.

Official setup guide (opens in a new tab) ↗
  1. Create a token

    In Settings → API tokens, create a token just for this assistant. Choose Read-only if it should only look.

    Sign in to create a token
  2. Add the server

    For you (all projects)
    ~/.cursor/mcp.json
    For one project
    .cursor/mcp.json
    Config file, token read from an environment variable ~/.cursor/mcp.json
    {
        "mcpServers": {
            "streamrepeater": {
                "url": "https://streamrepeater.com/mcp",
                "headers": {
                    "Authorization": "Bearer ${env:STREAMREPEATER_TOKEN}"
                }
            }
        }
    }
    Config file ~/.cursor/mcp.json
    {
        "mcpServers": {
            "streamrepeater": {
                "url": "https://streamrepeater.com/mcp",
                "headers": {
                    "Authorization": "Bearer YOUR_TOKEN"
                }
            }
        }
    }
  3. Check it works

    Open Cursor's settings, find MCP and check Stream Repeater is switched on and lists its tools.

    Then ask: List my channels and tell me which are live.

  • The first snippet reads the token from the STREAMREPEATER_TOKEN environment variable. Set it before you start Cursor.
  • Cursor asks you to approve each tool call unless you allow it.

VS Code (GitHub Copilot)

GitHub Copilot's agent mode in Visual Studio Code.

Official setup guide (opens in a new tab) ↗
  1. Create a token

    In Settings → API tokens, create a token just for this assistant. Choose Read-only if it should only look.

    Sign in to create a token
  2. Add the server

    For one project
    .vscode/mcp.json
    For you (all projects)
    MCP: Open User Configuration
    Config file .vscode/mcp.json
    {
        "inputs": [
            {
                "type": "promptString",
                "id": "streamrepeater-token",
                "description": "Stream Repeater API token",
                "password": true
            }
        ],
        "servers": {
            "streamrepeater": {
                "type": "http",
                "url": "https://streamrepeater.com/mcp",
                "headers": {
                    "Authorization": "Bearer ${input:streamrepeater-token}"
                }
            }
        }
    }
  3. Check it works

    Run MCP: List Servers from the Command Palette, then start Stream Repeater. Use Copilot Chat in Agent mode.

    Then ask: List my channels and tell me which are live.

  • VS Code asks for the token the first time the server starts and keeps it in its secret storage, so it is not saved in the file.
  • Note the top-level key is servers, not mcpServers.

Windsurf

The Windsurf editor and its Cascade agent.

Official setup guide (opens in a new tab) ↗
  1. Create a token

    In Settings → API tokens, create a token just for this assistant. Choose Read-only if it should only look.

    Sign in to create a token
  2. Add the server

    macOS and Linux
    ~/.codeium/windsurf/mcp_config.json
    macOS and Linux
    ~/.config/devin/mcp_config.json
    Windows
    %AppData%\devin\mcp_config.json
    Config file, token read from an environment variable mcp_config.json
    {
        "mcpServers": {
            "streamrepeater": {
                "serverUrl": "https://streamrepeater.com/mcp",
                "headers": {
                    "Authorization": "Bearer ${env:STREAMREPEATER_TOKEN}"
                }
            }
        }
    }
  3. Check it works

    In Cascade, open the MCP panel, press refresh, and check Stream Repeater lists its tools.

    Then ask: List my channels and tell me which are live.

  • Windsurf reads the token from the STREAMREPEATER_TOKEN environment variable. Set it before you start the editor.
  • Newer versions keep the file in the devin folder shown above. Open it from Cascade's MCP settings if you are not sure which one yours uses.

Zed

The fast, collaborative code editor.

Official setup guide (opens in a new tab) ↗
  1. Create a token

    In Settings → API tokens, create a token just for this assistant. Choose Read-only if it should only look.

    Sign in to create a token
  2. Add the server

    macOS and Linux
    ~/.config/zed/settings.json
    Open it from
    zed: open settings file
    Config file settings.json
    {
        "context_servers": {
            "streamrepeater": {
                "url": "https://streamrepeater.com/mcp",
                "headers": {
                    "Authorization": "Bearer YOUR_TOKEN"
                }
            }
        }
    }
  3. Check it works

    Open Settings → AI → MCP Servers. A green dot next to Stream Repeater means it is connected.

    Then ask: List my channels and tell me which are live.

  • You can also add it from Settings → AI → MCP Servers without editing the file. Zed only offers an OAuth sign-in when no Authorization header is set.

Cline

The open-source coding agent for VS Code, JetBrains and the terminal.

Official setup guide (opens in a new tab) ↗
  1. Create a token

    In Settings → API tokens, create a token just for this assistant. Choose Read-only if it should only look.

    Sign in to create a token
  2. Add the server

    Open it from
    MCP Servers → Configure → Configure MCP Servers
    For you (all projects)
    ~/.cline/data/settings/cline_mcp_settings.json
    One command
    cline mcp add streamrepeater https://streamrepeater.com/mcp --transport http \
      --header "Authorization: Bearer YOUR_TOKEN" --yes
    Config file cline_mcp_settings.json
    {
        "mcpServers": {
            "streamrepeater": {
                "type": "streamableHttp",
                "url": "https://streamrepeater.com/mcp",
                "headers": {
                    "Authorization": "Bearer YOUR_TOKEN"
                },
                "disabled": false,
                "autoApprove": []
            }
        }
    }
  3. Check it works

    Open the MCP Servers panel and check Stream Repeater shows as connected.

    Then ask: List my channels and tell me which are live.

  • Keep "type": "streamableHttp": without it Cline falls back to the older SSE transport.

Continue

The open-source assistant for VS Code and JetBrains.

Official setup guide (opens in a new tab) ↗
  1. Create a token

    In Settings → API tokens, create a token just for this assistant. Choose Read-only if it should only look.

    Sign in to create a token
  2. Add the server

    For one project
    .continue/mcpServers/streamrepeater.yaml
    Config file .continue/mcpServers/streamrepeater.yaml
    name: Stream Repeater
    version: 0.0.1
    schema: v1
    mcpServers:
      - name: streamrepeater
        type: streamable-http
        url: https://streamrepeater.com/mcp
        requestOptions:
          headers:
            Authorization: "Bearer YOUR_TOKEN"
  3. Check it works

    Reload Continue, switch to Agent mode and check the Stream Repeater tools appear.

    Then ask: List my channels and tell me which are live.

  • MCP tools only work in Continue's Agent mode.
  • Do not commit this file with a real token in it.

Goose

Block's open-source agent, on the desktop and the command line.

Official setup guide (opens in a new tab) ↗
  1. Create a token

    In Settings → API tokens, create a token just for this assistant. Choose Read-only if it should only look.

    Sign in to create a token
  2. Add the server

    macOS and Linux
    ~/.config/goose/config.yaml
    Config file, token read from an environment variable ~/.config/goose/config.yaml
    extensions:
      streamrepeater:
        enabled: true
        type: streamable_http
        name: streamrepeater
        uri: https://streamrepeater.com/mcp
        headers:
          Authorization: "Bearer $STREAMREPEATER_TOKEN"
        timeout: 300
  3. Check it works

    Restart Goose and ask it which Stream Repeater tools it has.

    Then ask: List my channels and tell me which are live.

  • The key is uri, not url. Goose replaces $STREAMREPEATER_TOKEN with the environment variable, so set it before you start Goose.
  • You can also run goose configure, choose Add Extension, then Remote Extension (Streamable HTTP), and add the header afterwards in the file.

ChatGPT

Add Stream Repeater to ChatGPT as a custom MCP server and sign in. There is no token to paste.

Official setup guide (opens in a new tab) ↗
  1. Add a custom connector

    1. Open chatgpt.com/plugins, click +, then Add custom MCP server.
    2. Enter Stream Repeater as the name, and the server URL below under Connection.
    3. Choose OAuth for authentication, confirm the warning, then click Create as a plugin.
    4. Sign in to Stream Repeater when asked, choose a workspace, and choose whether ChatGPT can make changes.
    Server URL
    https://streamrepeater.com/mcp
  2. Check it works

    Check the Stream Repeater tools are listed, then use the plugin in a chat.

    Then ask: List my channels and tell me which are live.

  • OpenAI decides which ChatGPT plans can add custom MCP servers, and a workspace admin may need to allow it. If you do not see Add custom MCP server, your plan or workspace does not offer it.
  • You can see and disconnect ChatGPT in Settings → Connected apps.

Other MCP clients

Any client that supports the Streamable HTTP transport and can send an Authorization header.

Official setup guide (opens in a new tab) ↗
  1. Create a token

    In Settings → API tokens, create a token just for this assistant. Choose Read-only if it should only look.

    Sign in to create a token
  2. Add the server

    Config file
    {
        "mcpServers": {
            "streamrepeater": {
                "type": "http",
                "url": "https://streamrepeater.com/mcp",
                "headers": {
                    "Authorization": "Bearer YOUR_TOKEN"
                }
            }
        }
    }
    Test the connection with curl
    curl -X POST https://streamrepeater.com/mcp \
      -H "Authorization: Bearer YOUR_TOKEN" \
      -H "Content-Type: application/json" \
      -H "Accept: application/json, text/event-stream" \
      -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"curl","version":"1"}}}'
  3. Check it works

    Send the initialize request with curl. A JSON reply with serverInfo means the URL and token are right; 401 means the token is wrong.

    Then ask: List my channels and tell me which are live.

  • If your client can only start local (stdio) servers, the open-source mcp-remote bridge can connect it: npx -y mcp-remote SERVER_URL --header "Authorization: Bearer YOUR_TOKEN".

https://streamrepeater.com/mcpFull setup guide and the list of tools →

Works with the assistant you already use

Claude and ChatGPT sign in with OAuth. Any other MCP client that supports the Streamable HTTP transport connects with an API token. Choose one to see its setup.

Claude and ChatGPT sign in with OAuth: add the server URL as a custom connector, sign in to Stream Repeater, then choose a workspace and whether the assistant can make changes. There is no token to paste, and you can disconnect either one in Settings → Connected apps. In ChatGPT, OpenAI decides which plans can add custom MCP servers.

Prefer to write the code yourself?

The assistant uses the same public API you can call from your own scripts and services.

REST API

159 operations covering channels, destinations, recordings, clips, chat, analytics and Podcast Rooms, described in OpenAPI 3.1.

curl https://streamrepeater.com/api/v1/channels \
  -H "Authorization: Bearer $TOKEN"

Webhooks

Get a signed POST when something happens, retried up to 6 times if your endpoint is down. Events include:

  • stream.live
  • stream.offline
  • destination.failed
  • recording.started
  • bandwidth.warning
  • podcast.room.started
  • podcast.room.ended
  • podcast.guest.joined
  • podcast.guest.left

For AI crawlers too

A plain-text summary of the product, the API and every MCP tool, for models that read the web.

Questions

What is MCP?

The Model Context Protocol is an open standard that lets AI assistants use tools in other apps. Stream Repeater runs an MCP server, so an assistant that supports MCP can work with your streams when you ask it to.

Which assistants work with it?

Claude (on the web, the desktop and your phone), ChatGPT, Claude Code, Codex, Cursor, VS Code and any other MCP client that supports the Streamable HTTP transport. Claude and ChatGPT sign in with OAuth; the others use an API token. The setup page has step-by-step instructions for each.

Does it work in ChatGPT or on claude.ai?

Yes. Add Stream Repeater as a custom connector in Claude, or as a custom MCP server in ChatGPT, then sign in to Stream Repeater and choose a workspace and whether the assistant can make changes. There is no token to paste, and you can disconnect it any time in Settings → Connected apps. In ChatGPT, OpenAI decides which plans can add custom MCP servers.

Can the assistant see my stream key?

No. Stream keys, destination URLs and passwords are never returned to an assistant. It gets masked connection details and is told where you can copy the real ones.

Can I let it look but not touch?

Yes. Create a Read-only token. The assistant can then check health, usage, recordings and chat, but every tool that changes something is refused.

How do I know what it changed?

Every change made through MCP is written to your workspace Activity log with the tool, the token and the person it belongs to.

Does it cost extra?

No. API tokens and the MCP server come with every plan. What a tool can do still depends on your plan: recording tools need a plan with recording, for example.

Does the assistant run all the time?

No. It only does something when you ask it to, in your assistant. Stream Repeater keeps relaying your streams by itself either way.

Give your assistant the keys to the control room, not the stream keys

Create a free workspace, make an API token and connect your assistant.

14 days of Pro free, no card Start for free