MCP for LLMs and AI agents

Plug your AI agents straight into your data

Viral Manager exposes an MCP server with 50 tools. Claude, or any MCP-capable agent, can read your discovery feed, your assignments and your library, and act on them.

Start Free Trial

7-day free trial, cancel anytime

MCP server50 tools, read and write. Your assistant works inside your workspace. Connected
list_viral_postsniche: fitness, min_score: 80
get_post_analysispost: lea.fit, 412K views
create_assignmentmodel: Maya, due: Friday
list_watchlist_accountssort: growth, limit: 20
get_trending_soundsniche: fitness, window: 7d
summarize_my_accountsperiod: yesterday

Claude, connected to your workspace

50 tools, read and write

List viral posts, read an AI analysis, create an assignment, tag a batch of posts, move folders, pull a download link. Not a read-only feed: an agent can run part of your operation.

What an agent can actually do

Most tools stop at an export button. Viral Manager exposes its actual operations to AI agents through MCP, so an agent can read the night's viral posts, draft the briefs and assign them before your first coffee.

Connecting an agent

  1. 1

    Generate a key

    From your settings, or connect via OAuth from an MCP client.

  2. 2

    Point it at the server

    One endpoint, standard MCP. Claude Desktop, Claude Code and others connect as-is.

  3. 3

    Ask in plain language

    The agent picks the right tools. You describe the outcome, not the API calls.

  4. 4

    Let it run daily

    Schedule it and the routine happens without you.

Start Free Trial

Where the server is listed

The server is published in the official MCP registry, which most directories mirror automatically. If your client browses a catalogue rather than a URL, it finds Viral Manager there.

Official MCP registryPublished as com.viral-managers/viral-manager. The reference registry, and the source downstream directories pull from.GlamaConnector listing with the full tool catalogue and a reachability check on the endpoint.SmitheryServer listing with connection details for clients that install from a directory.

Frequently asked questions

Which plans include MCP?

Scale and Empire. It is also available on white label workspaces.

Which clients are supported?

Any MCP-compatible client. Claude is the most common, but the protocol is open and the server is standard.

Is my data used to train a model?

No. The server exposes your workspace to the agent you connect. What that agent's provider does is governed by your agreement with them, not by us.

Can an agent delete things?

Only what the key's permissions allow. Destructive tools respect the same role gating as the interface.

Is there a rate limit?

Yes, sized so a normal automation routine never hits it.

Do I need to write code?

No. Connect an MCP client, then ask in plain language. Writing scripts against the API keys is possible but not required.

Can I restrict what an agent can reach?

Yes. An agent inherits the permissions of the key or account it connects with, including category restrictions on team members.

Which operations are exposed?

Discovery and viral posts, analyses, assignments, folders, tags and categories, assets and download links, watchlist accounts, and your own account analytics.

Still have a question? Contact us

Keep reading

Viral DiscoveryThe feed an agent reads first.Production trackingWhat an agent can write into.White labelMCP on your own branded workspace.

Ready to stop guessing?

Start Free Trial

7-day free trial, cancel anytime