On this page
TokPortal is an MCP server for social media posting that lets AI agents publish TikTok, Instagram, and YouTube videos through real physical devices, local SIM cards, and human operators. It gives Claude or custom agents a tool interface for native in-app posting, campaign control, approvals, and webhooks.
TokPortal is programmable, organic social-media distribution infrastructure — The Human API. The MCP server gives agents a safe tool layer for creating posting jobs, selecting countries, attaching assets, requesting approval, and receiving completion events without forcing every workflow through custom REST code.
This is built for teams that already generate content: AI video tools, agencies, app-growth teams, labels, e-commerce operators, and technical marketers. If you are looking for consumer utilities like “download video tiktok with watermark,” “tikip us,” or “rtx fury,” this is not that workflow; this page is for publishing approved brand videos through real devices. For agent-specific setup details, start with TokPortal MCP for AI agents or open the TokPortal developer documentation.
Use MCP to post TikTok videos
To use MCP to post TikTok videos, connect your agent to TokPortal’s MCP server, expose the posting tools your workspace allows, then let the agent create a campaign job with a video file, caption, target country, account group, schedule, and approval state.
The important difference is execution. TokPortal posts inside the real TikTok app on physical smartphones with local SIM cards in 20+ countries. That means native app features such as sounds, location tags, and in-app editing are available in workflows where a standard developer endpoint may not cover the full publishing experience. For the native sound path, read how TokPortal handles TikTok sounds through native in-app posting.
- Agent input: video URL, caption, country, account pool, posting window, campaign label.
- TokPortal action: assign a real account and real device, queue the post, route operator review where required.
- Agent output: post URL, status, analytics event, webhook payload, and retry reason if the job needs correction.
Claude MCP integration with social
A Claude MCP social media workflow works best when Claude is the planner and TokPortal is the execution layer. Claude can inspect a campaign brief, choose a country, rewrite captions for local context, check asset readiness, and call TokPortal tools only after the campaign rules are satisfied.
In practice, the MCP server should expose narrow tools, not unlimited workspace authority. Common tool names include create_post_job, list_account_groups, check_campaign_status, request_operator_review, and get_post_results. Claude then calls those tools through the Model Context Protocol instead of inventing its own API calls.
For a deeper walkthrough of the agent pattern, use the AI agent TikTok MCP setup guide. If your stack is workflow-first rather than chat-first, compare it with TokPortal + n8n automation.
Build an AI agent that posts to Instagram
Define the agent’s posting contract
Give the agent a fixed schema: video asset, caption, platform, country, account group, schedule, approval requirement, and campaign tag. Do not let the agent create free-form posting requests.
Connect the agent to TokPortal MCP
Install the MCP server in the agent runtime, add the TokPortal workspace token, and allow only the tools needed for campaign creation, status checks, and result retrieval.
Add a human approval checkpoint
For Instagram Reels, route the final caption, video, cover, and posting window through approval before publishing. TokPortal is human-in-the-loop infrastructure, so approval should be explicit.
Select the country and account group
Brands often buy country-specific distribution. Match the campaign to the right account pool, local device, and posting window before the job is queued.
Publish through the native app path
TokPortal executes the job through real devices and native app posting where supported, then returns the post URL and status event.
Close the loop with webhooks
Send completion, failure, review, and analytics events back to your agent or database. Use webhook events to make the next agent decision data-driven.
MCP server vs REST API for posting
Feature
TokPortal MCP server
TokPortal REST API
Best user
Interaction style
Control model
Best workflow
When not to use
Use MCP when the agent needs context: campaign brief, market choice, caption variants, asset readiness, or approval state. Use REST when the job is already decided and your system only needs to send structured requests at volume.
Most production teams use both. The agent uses MCP for reasoning and decisioning; the backend uses REST for deterministic execution, audit logs, and dashboard state. The REST implementation details live in the TokPortal REST API developer guide.
Connect TokPortal to Claude MCP
- Create a TokPortal workspace and generate a developer token.
- Install the TokPortal MCP server in the Claude Desktop or agent runtime configuration.
- Expose only the tools the agent needs: account lookup, campaign creation, post scheduling, status checks, and result retrieval.
- Set workspace-level approval rules for platforms, countries, budgets, and account groups.
- Test with one TikTok or Instagram job before enabling multi-account campaigns.
- Subscribe to webhook events so the agent can react to review, completion, and analytics updates.
Automate TikTok posting with agents
Agent-led TikTok automation should not mean letting a model publish whatever it wants. The durable pattern is a constrained agent: it can read the brief, choose from approved assets, pick a country, propose a caption, request review, create a TokPortal job, and monitor completion.
A typical AI video team generates 100 short clips from Sora, Runway, Kling, Veo, Captions, or Creatify, then needs distribution across markets. TokPortal becomes the post-generation layer: the agent selects the best clip, TokPortal assigns real-device execution, and webhooks push results back into the content database. For event design, use the TokPortal webhook events reference.
If you want visual orchestration around the agent, connect the same posting flow to TokPortal + Make.com or TokPortal + Zapier.
20+
countries with real-device distribution coverage
150,000+
accounts under TokPortal management
4,276
active business clients using TokPortal infrastructure
6B+
organic video views generated through TokPortal campaigns
Original implementation rule: keep the model out of final execution
Where TokPortal MCP fits
- AI video teams that need a post-generation distribution layer.
- Agencies coordinating client campaigns across TikTok, Instagram, and YouTube.
- Developers who want Claude or internal agents to manage campaign setup without writing every workflow as custom UI.
- Teams that need native in-app posting features rather than a narrow upload endpoint.
- Operators who want webhook feedback loops for agent decisions.
Where TokPortal MCP is not the answer
- Simple one-account scheduling where a basic social scheduler is enough.
- Consumer video downloading, watermark removal, or repost utility workflows.
- Campaigns without rights to the creative assets being published.
- Teams that are not ready to define approval rules before giving an agent tool access.
Build your agent posting workflow
Open the TokPortal developer docs, connect the MCP server, and launch a controlled test campaign through real-device social distribution.
Can an MCP server post to TikTok?+
How is TokPortal MCP different from the TikTok Content Posting API?+
Can Claude post to Instagram using TokPortal?+
Should I use MCP or the REST API?+
Does TokPortal support webhooks for agent feedback loops?+
What platforms can TokPortal agents post to?+

Written by
Vincent Tellenne
Co-founder & CEO
Vincent is a co-founder and CEO of TokPortal. He works on the infrastructure and operating model behind scaled organic social media distribution.
Learn more about this topic with AI
Related Resources
TokPortal MCP Server: Setup, GitHub, npm and Tools
Connect TokPortal MCP to Claude, Codex or Cursor. Find the official repo, browse 91 management tools and verify setup before creating accounts or posting.
TokPortal MCP Server: Let AI Agents Manage Your TikTok Distribution
Deploy TokPortal as an MCP server for AI agents. Let Claude, ChatGPT, or custom AI agents autonomously create TikTok accounts, post videos, and manage campaigns.
TokPortal + n8n: Automate Account Creation and Video Posting
Integrate TokPortal with n8n to automate TikTok account creation, video posting, and campaign management. Step-by-step guide with workflow templates for 2026.
How to Add TikTok Sounds via API: Native In-App Posting Explained
Learn how to add TikTok sounds via API using TokPortal's native in-app posting. Add trending sounds, control volume levels, and use carousels — all programmatically.
TokPortal REST API: Complete Developer Guide
The complete developer guide to the TokPortal REST API. Learn how to create TikTok and Instagram accounts, upload videos with native sounds, manage warming, and build automation pipelines programmatically.
TokPortal Webhook Events: Complete Reference Guide
Complete reference guide to TokPortal webhook events. Learn every event type, payload structure, and how to connect real-time social media automation to n8n, Make, Zapier, or your own backend.
From the guide to your first campaign
You make the content.
TokPortal runs the account operations.
Create and operate TikTok and Instagram accounts in your target market. Local operators handle the account work and native publishing. You control the brief, content and campaign from the dashboard, API or MCP.
- 01Choose your market
- 02Define each account
- 03Schedule your content
Review market availability and pricing before ordering. No reach forecast is built into your plan.