TokPortal MCP Server for Cursor

The public MCP package is tokportal-mcp. It lets Cursor call generated TokPortal tools using your API key. This is a local stdio MCP server; your key is passed to the TokPortal API by the local process.

Install

Add this to .cursor/mcp.json:

{
  "mcpServers": {
    "tokportal": {
      "command": "npx",
      "args": ["-y", "tokportal-mcp"],
      "env": {
        "TOKPORTAL_API_KEY": "sk_your_key_here"
      }
    }
  }
}

Get your key from the Developer Portal.

Reference: MCP Server

Verify the connection

Ask Cursor:

"Use the TokPortal MCP tools to show my credit balance."

If connected, Cursor should call tokportal_get_credit_balance.

Create a bundle from Cursor

Use a prompt that forces Cursor to show the payload before executing:

"Prepare a TokPortal request for a US TikTok bundle with 10 video slots and Advanced Niche Warming on 6 SaaS/tech targets. Show the exact MCP tool call first. Wait for confirmation before running it."

Expected tool shape:

{
  "tool": "tokportal_create_bundle",
  "args": {
    "bundle_type": "account_and_videos",
    "platform": "tiktok",
    "country": "US",
    "videos_quantity": 10,
    "wants_advanced_warming": true,
    "advanced_warming_terms": [
      "saas tools",
      "b2b software",
      "productivity apps",
      "startup founder",
      "developer tools",
      "no code automation"
    ],
    "external_ref": "cursor-saas-us"
  }
}

wants_advanced_warming: true must travel with advanced_warming_terms (or advanced_warming_terms_count); on its own the target list is rejected with ADVANCED_WARMING_FLAG_REQUIRED. Targets are 3-30, in multiples of 3, each 2-50 characters. At the standard rate of 5 credits per target the 6 targets above are 30 credits, on top of 32 for the account and 2 per video slot: 32 + (10 x 2) + 30 = 82 credits.

Costs are returned by the API as credits_charged and cost_breakdown; do not hardcode them in your agent logic.

Configure videos from project files

Cursor can read local files if you allow it. A useful workflow is:

"Read marketing/video-plan.md, extract 5 captions and video URLs, then configure positions 1-5 on bundle bnd_.... Use target_publish_date starting 5 days from today, at most 3 videos on any single day. Show the batch payload before executing."

The earliest target_publish_date a bundle accepts is today + 3 days while its account is still being created, and today + 1 once the account is delivered or when the bundle runs on an existing account. A 4th video on the same day is rejected with VIDEOS_PER_DAY_EXCEEDED.

Expected API surface:

tokportal_batch_configure_bundle_videos
PUT /bundles/{id}/videos/batch
body shape: { "videos": [...] }

Reference: Configure Videos

Monitor status and analytics

Useful tools:

Tool	Use
`tokportal_get_bundle`	Inspect bundle status and configured fields
`tokportal_get_bundle_publish_readiness`	See why a bundle cannot publish yet
`tokportal_list_accounts`	List delivered accounts
`tokportal_get_account_analytics`	Fetch account analytics when available
`tokportal_list_account_video_analytics`	Fetch post-level analytics

Analytics availability depends on the account/post state and your plan. See Analytics.

Operational guardrails

Ask Cursor to confirm mutating calls before execution.
Use external_ref to connect TokPortal records to your internal campaign IDs.
Do not ask the agent to promise views, engagement, reach, or platform enforcement outcomes.
Use Webhooks for production monitoring instead of relying only on chat output.