If you came here looking for the viral.app TikTok MCP server, here is the direct answer: the MCP is deprecated.
We keep the old context around for transparency, because older links, client configs, and search results still point to it. But new work should start with the viral.app API or the viral.app agent skill.
It is an upgrade.
The old MCP answered a narrow question: "Can my AI client fetch TikTok data?" The API and agent skill answer the more useful question: "Can my agent work with the same short-form analytics system my team uses in the dashboard?"
The answer is yes. TikTok, Instagram, YouTube, and Facebook are all exposed through the API platform parameter. That means TikTok analytics API, Instagram Reels analytics API, YouTube Shorts analytics API, and Facebook Reels tracking workflows can live in one integration instead of four disconnected experiments.
For a broader buying guide, read our social media analytics API guide. This article is narrower: it explains what replaced the TikTok MCP and how to think about the new API and skill workflow.
The old page was a free-tool landing page for a TikTok MCP server. It taught users how to connect MCP clients such as Cursor, Raycast, Claude Desktop, Windsurf, and Cline to a hosted MCP endpoint.
That made sense when the product surface was smaller. It makes less sense now.
viral.app now has a real API surface with interactive docs, API keys, tracked analytics, live platform lookups, projects, Creator Hub resources, payout workflows, exports, and an agent skill that can call that functionality from a local CLI. The MCP became the smaller path.
The first MCP exposed a small set of TikTok lookup tools for AI clients.
The API now covers live data, tracked accounts, tracked videos, analytics, projects, Creator Hub, payouts, CSV exports, and video downloads.
The viral.app skill lets Codex, Open Claw, Claude Code, Claude Cowork, and other CLI-capable agents use the API with reusable workflows.
MCP still has a place. The official MCP documentation describes a client-server model for connecting AI apps to external tools and data. That model is useful when a tool only needs a small, fixed set of actions.
viral.app has moved past that shape. The API is the stable surface. The agent skill is the ergonomic layer on top.
The old TikTok MCP was intentionally limited. It could look up a TikTok profile, inspect a TikTok video, list profile videos, and answer simple TikTok analytics questions.
The API does much more:
Search users, videos, and hashtags. Fetch live account details, video details, account videos, and video comments across supported platforms.
Read tracked accounts, tracked videos, top videos, top accounts, KPIs, metric history, posting activity, and exports.
Add tracked accounts, mark competitor accounts, bulk update competitor flags, and keep internal versus competitor reporting clean.
List creators, assign campaigns, inspect Creator Hub overview tables, calculate payouts, initiate payouts, and review payment windows.
Manage projects, associate accounts with projects, export accounts or videos, and wire the results into client reports or internal BI.
Use API keys, OpenAPI docs, a local CLI, and the viral.app skill so agents can ask grounded questions and return useful reports.
In practice, the replacement is larger than "TikTok MCP with a different install command." It is a larger product surface.
You can ask an agent to:
Most MCP servers stop at "tool call returns data." The viral.app API reaches into the same operating system your team uses for dashboards, projects, campaigns, and payout workflows.
This part matters commercially.
Many social data vendors treat API access as an enterprise feature. You may need a separate contract, a higher minimum, a paid add-on, or a custom data package before you can automate anything.
viral.app takes a simpler route: API access is included with every plan tier starting at Ultra. If your team is already on the plan tier built for heavier creator operations, the programmable layer is part of the product.
If you are comparing TikTok analytics API pricing, YouTube Shorts analytics API access, Instagram Reels reporting APIs, or Facebook Reels tracking tools, do not compare only the endpoint list. Check whether API access is included in the plan you would actually buy.
The viral.app agent skill is the cleaner replacement for MCP-style usage.
It is built for agents that can read a skill, run a CLI, and keep work grounded in your organization
through VIRAL_API_KEY. That includes the obvious developer agents, but the pattern is wider than
one client.
Install options:
clawhub install viral-app
npx skills add https://github.com/fmd-labs/viral-app-skills --skill viral-app
Then create an API key in API Keys, expose it as VIRAL_API_KEY in the same
environment where the agent runs, and start with a read-only analytics prompt.
The official Claude Code skills documentation, OpenAI Codex help, and skills.sh CLI docs are useful if you are standardizing skills across multiple agents.
Start in the interactive API docs. The docs expose request parameters, schemas, and example calls, so you do not need to guess the shape of the integration.
These are the most useful starting points if you are migrating from a TikTok MCP mindset:
That endpoint set is why the API works for more than "TikTok analytics API" searches. It also fits:
The old MCP endpoint was:
https://viral.app/api/mcp
The old server focused on TikTok-only tools such as profile lookup, video lookup, recent profile videos, and simple natural-language analysis around those retrieved objects.
That context is useful if you are cleaning up an old Cursor, Raycast, Claude Desktop, Windsurf, or Cline config. It is not the recommended path for new builds.
If you still have old MCP configuration that calls mcp-remote against https://viral.app/api/mcp,
treat it as legacy config. Move the workflow to API keys, the API docs, and the
agent skill.
Audit Cursor, Raycast, Claude Desktop, Windsurf, Cline, or other client configs for the old viral.app MCP endpoint.
Generate an org-scoped key in the dashboard and provide it to your agent as VIRAL_API_KEY.
Use ClawHub for Open Claw or the skills CLI for Codex, Claude Code, and compatible skill-based agents.
Ask for a leaderboard, KPI comparison, live account lookup, or tracked video summary before allowing write actions.
A good first prompt:
Generate a weekly viral.app leaderboard for the latest rolling 7-day window.
Use tracked account and tracked video analytics from this org. Return the top
10 accounts by views gained, the top 10 accounts by engagement rate, and the
top 10 videos across TikTok, Instagram, YouTube, and Facebook if available.
Include direct viral.app links.
Try without risk with our 7-day free trial.
Go based on data, not by luck.
Pay everyone on your own terms.
Eliminate hours of manual work.
Pay $0 today · Cancel anytime
