Remote MCP server

Give your AI a safe way to publish.

Connect Codex, Claude Code, Cursor, or another MCP client to one reviewed workflow for drafts, schedules, media, and publishing across nine social platforms.

Remote server ready

Let your AI connect Solnk.

Copy one instruction into Codex, Claude Code, Cursor, or another MCP-capable agent. It will choose the correct configuration method, ask for your key, and verify the tools.

Set up Solnk as a remote MCP server in the AI client I am using.

https://mcp.solnk.com/mcp

Create an API key

Tools that feel native to the conversation.

Your agent calls the right Solnk tool while you stay in the client where the work already happens.

solnk_publish

Create a draft, schedule a post, or publish immediately.

solnk_confirm_publish

Release a reviewed draft immediately or schedule it.

solnk_get_publish_status

Check aggregate and per-destination delivery status.

solnk_list_accounts

See which social accounts are available to the agent.

solnk_create_media_upload

Start a local image or video upload for publishing.

solnk_list_teams

List owned teams and identify the selected default team.

solnk_set_default_team

Choose which owned team is used as the default fallback.

solnk_delete_team

Delete an extra team with explicit confirmation and content migration.

The agent prepares a draftPrepared
You review it in SolnkYour decision
Solnk publishes to selected accountsDelivered

Let it draft. Keep the last word.

Start with draft creation and human approval. Direct publishing stays available when the workflow has earned your trust, while scoped API keys control what each agent can access.

Create a scoped API key ↗

How to turn an AI draft into a reviewed social post

Start with a connected account and paid-plan MCP access. Choose your destinations, review the draft, and check the result for every account after publishing.

  1. Choose the exact destinations

    Ask the agent to call solnk_list_accounts and show the account names. Select each destination explicitly, especially when several client accounts share a platform.

  2. Create a draft and stop for review

    Use solnk_publish with publish_mode: "draft" and the selected account IDs in targets. Keep the returned publish ID. Review final text, links, media order, and each destination in Solnk.

    Prepare a draft for my selected accounts. Show each destination, final text, and attached media. Do not confirm, schedule, or publish until I approve this exact version.
  3. Choose whether approval must be enforced

    A prompt asking the agent to wait does not revoke its publishing permission. For server-enforced approval, enable team approval and use a member credential. Publishing or confirming then returns approval_status: "pending_approval". An owner or admin must approve; owner/admin credentials are exempt, and an unconfirmed draft has not entered the approval gate.

  4. Release the reviewed draft and verify delivery

    After your review, call solnk_confirm_publish for that draft, optionally with a future UTC scheduled_at. If team approval is pending, an owner/admin must approve through POST /api/v1/publishes/{id}/approve before delivery. Then use solnk_get_publish_status and inspect every target.

    publishing and processing are unfinished. Aggregate published means every target has a platform content ID; partial means some succeeded and some failed. Keep destination URLs and check approval_status separately from lifecycle status.

Check the current API reference · Build an n8n scheduling workflow

Configure it manually when you need control.

Native remote-MCP clients can use the server URL and Bearer key directly. Other clients can connect through the localmcp-remotebridge.

Read the API documentation

Codex

Add this to ~/.codex/config.toml, then run codex mcp login solnk.

[mcp_servers.solnk]
url = "https://mcp.solnk.com/mcp"
auth = "oauth"
default_tools_approval_mode = "writes"

Cursor / Claude Code

This form avoids the known Windows argument-spacing problem.

{
  "mcpServers": {
    "solnk": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote@latest",
        "https://mcp.solnk.com/mcp",
        "--header",
        "Authorization:${SOLNK_AUTH_HEADER}"
      ],
      "env": {
        "SOLNK_AUTH_HEADER": "Bearer sk_..."
      }
    }
  }
}

ChatGPT web

Open Plugins or developer mode, add the public endpoint, and complete OAuth. Access depends on your ChatGPT account and workspace policy.

https://mcp.solnk.com/mcp

MCP or the REST API?

Use MCP

Choose MCP when you want Codex, Claude Code, Cursor, or another AI client to call publishing tools inside the conversation.

Use the REST API

Choose the API for your own product, backend service, n8n flow, or scheduled automation. Explore the AI agent API.

MCP FAQ

It is a hosted Model Context Protocol server that gives an MCP-compatible AI agent Solnk publishing tools. The agent can create drafts, schedule posts, publish, list connected accounts, and attach media without custom API wiring.

Codex connects directly over Streamable HTTP with OAuth or a Bearer key. Cursor and Claude Code can connect directly when their version supports remote MCP, or through mcp-remote. ChatGPT web uses plugins or developer-mode connections, and availability depends on the account and workspace policy.

Use https://mcp.solnk.com/mcp and follow the client-specific configuration on this page. Prefer OAuth when the client supports it. Use a scoped Solnk API key only for clients that require manual Bearer authentication.

A draft-first prompt asks the agent to wait; it does not remove publishing permission. For server-enforced review, enable team approval and use a member credential. Owner/admin credentials are exempt. Check approval_status separately from delivery status.

Use MCP when an AI client should call publishing tools for you. Use the REST API when you are building your own product, backend integration, or automation. Both reach the same publishing layer.

MCP and API access are included with paid plans. You can start with a free account, connect your social accounts, and prepare the workflow before upgrading.

Give your agent the connection. Keep control of the publish.

Create an account, connect your social profiles, and let your AI prepare the first reviewed draft.