Slidestorm MCP Quickstart
Connect your agent to Slidestorm in minutes and run your first tool call.
https://mcp.slidestorm.ai/mcpClients that support OAuth (ChatGPT, Claude, Claude Code, Cursor) only need this URL. When you connect, a Slidestorm page opens asking for your API token, which is stored securely on the connection. Other clients can send the token directly in an Authorization header. Tokens don't expire; you can revoke them at any time from Settings → API Tokens.
Once connected, ask your agent in plain language: “Create a 6-slide slideshow about X, wait for it to finish, and give me the link.” Chat apps can't show the slide images inline, so the agent replies with a link to view and edit the slideshow in Slidestorm, and can then post it to your connected TikTok account.
Requires a ChatGPT Plus, Pro, Business, Enterprise, or Edu plan. Custom connectors are added on the web app.
1) In ChatGPT open Settings → Apps (or Connectors) → Advanced settings and turn on Developer mode.
2) Click Create (or Add custom connector).
Name: Slidestorm
MCP server URL: https://mcp.slidestorm.ai/mcp
Authentication: OAuth
3) Click Create, then Connect. A Slidestorm page opens asking for your API token.
4) Paste a token from Settings → API Tokens and click Connect.
5) Start a new chat, enable Slidestorm from the tools menu, and ask it to run check_usage.
ChatGPT may ask you to approve each Slidestorm action the first time. That is normal for custom connectors.1) In Claude open Settings → Connectors → Add custom connector.
Name: Slidestorm
URL: https://mcp.slidestorm.ai/mcp
2) Click Add, then Connect. A Slidestorm page opens asking for your API token.
3) Paste a token from Settings → API Tokens and click Connect.claude mcp add --transport http slidestorm https://mcp.slidestorm.ai/mcp
# Then inside Claude Code run /mcp and choose Slidestorm → Authenticate.Cursor supports OAuth for remote servers, so the URL alone is enough. Cursor will prompt you to sign in on first use.
{
"mcpServers": {
"slidestorm": {
"url": "https://mcp.slidestorm.ai/mcp"
}
}
}Prefer to skip the sign-in page? Send your token directly instead:
{
"mcpServers": {
"slidestorm": {
"url": "https://mcp.slidestorm.ai/mcp",
"headers": {
"Authorization": "Bearer YOUR_SLIDESTORM_TOKEN"
}
}
}
}Use the Claude connector steps above if custom connectors are available on your plan. Otherwise, add the server via mcp-remote:
{
"mcpServers": {
"slidestorm": {
"command": "npx",
"args": [
"-y",
"mcp-remote@latest",
"https://mcp.slidestorm.ai/mcp",
"--header",
"Authorization:Bearer YOUR_SLIDESTORM_TOKEN"
]
}
}
}{
"mcp": {
"servers": {
"slidestorm": {
"url": "https://mcp.slidestorm.ai/mcp",
"transport": "streamable-http",
"headers": {
"Authorization": "Bearer YOUR_SLIDESTORM_TOKEN"
}
}
}
}
}First-run verification
Ask your agent to run these tools in order. If each succeeds, your MCP setup is complete.
1) "Use Slidestorm to run check_usage."
2) "Use Slidestorm to run list_accounts."
3) "Create a 3-slide slideshow about coffee brewing tips, wait for it to finish, and give me the link."TikTok publish behavior
MCP posting uses TikTok MEDIA_UPLOAD mode. That means posts are uploaded to TikTok inbox first, then you must open TikTok app and publish manually.
When using create_post or update_post:
- Set platform_configurations.tiktok.post_mode to MEDIA_UPLOAD (or omit it).
- Do not expect automatic public publishing.
- Keep text within TikTok limits (rejected otherwise):
- title: 90 characters (derived from caption start)
- description: 4000 characters
- emoji/symbol caveat: TikTok limits are validated in UTF-16 runes, so some emoji count as more than one character
- After upload succeeds, open TikTok app and publish manually.Common errors and fixes
Authorization: Bearer YOUR_SLIDESTORM_TOKEN to your config.check_usage and top up before creating slideshows.wait_for_slideshow rather than polling in a loop.Next action
Continue with the API workflow guide for end-to-end examples (create slideshow → poll → post).