Publora MCP Server
Control Publora directly from AI assistants like Claude Code, Claude Desktop, Cursor, and any MCP-compatible client. No coding required -- just describe what you want in plain English.
Overview
Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and services. Publora provides a remote MCP server at mcp.publora.com -- one-time setup, then just talk to your AI.
Perfect for:
- Marketers -- schedule campaigns, check analytics, manage multiple accounts
- Content creators -- post to all platforms at once without leaving your AI chat
- Business owners -- delegate social media tasks to your AI assistant
- Developers -- integrate Publora into AI-powered workflows
Just describe what you want:
"Show my connected social accounts" "Schedule a post to LinkedIn for tomorrow at 9am" "Delete my scheduled post for tomorrow" "Post this announcement to all my accounts"
Quick Start
Using Claude? Start here
Publora is in the Claude connectors directory. Open claude.ai/directory/publora and select Connect — that is the whole setup, on web, desktop and mobile alike. The steps below cover other MCP clients and Claude Code, which can sign in with OAuth or use a static key.
1. Get Your API Key
Log in to Publora → API in the sidebar → Copy your key (sk_...). Click MCP in the sidebar for additional setup instructions.
2. Configure Your MCP Client
Add Publora to your MCP configuration:
{
"mcpServers": {
"publora": {
"type": "http",
"url": "https://mcp.publora.com",
"headers": {
"Authorization": "Bearer sk_YOUR_API_KEY"
}
}
}
}Note: Both
https://mcp.publora.com/andhttps://mcp.publora.com/mcpwork as the MCP endpoint URL.
3. Restart Your Client
After restarting, your client lists 20 personal tools + 12 Agency tools (32 total). Personal tools are included on every plan, with individual features subject to plan limits; Agency tools require an Agency (Company) account. Try asking:
"List my connected social media accounts"
Client-Specific Setup
Claude Code (CLI)
Sign in with OAuth — no API key:
claude mcp add --transport http --scope user publora https://mcp.publora.com/mcpThen run /mcp, select publora, choose Authenticate and sign in to Publora in the browser.
To use a static key instead (CI, headless machines), edit ~/.claude.json or create .mcp.json in your project:
{
"mcpServers": {
"publora": {
"type": "http",
"url": "https://mcp.publora.com/mcp",
"headers": {
"Authorization": "Bearer sk_YOUR_API_KEY"
}
}
}
}Restart Claude Code. Verify with /mcp command.
Claude Desktop
Claude Desktop uses the same connectors as Claude on the web: open claude.ai/directory/publora and select Connect. There is no config file to edit — Settings → Developer → Edit Config is for local MCP servers only. To add Publora by URL instead, see MCP Client Setup.
Cursor
Create .cursor/mcp.json in your project:
{
"mcpServers": {
"publora": {
"type": "http",
"url": "https://mcp.publora.com",
"headers": {
"Authorization": "Bearer sk_YOUR_API_KEY"
}
}
}
}Using with Other MCP Servers
MCP servers don't conflict -- Claude loads all servers and merges their tools. Example with Context7:
{
"mcpServers": {
"context7": {
"command": "npx",
"args": ["-y", "@upstash/context7-mcp"]
},
"publora": {
"type": "http",
"url": "https://mcp.publora.com",
"headers": {
"Authorization": "Bearer sk_YOUR_API_KEY"
}
}
}
}Available Tools (32 total)
Publora exposes 20 personal tools + 12 Agency tools (32 total). The tables below list the personal tools. The 12 company_* tools are for Agency client work and only work for Agency (Company) accounts.
Posts
| Tool | Description |
|---|---|
list_posts |
List posts with filters (status, platform, dates, pagination; responseFormat: concise for token-lean previews) |
create_post |
Create a draft, schedule a post, or publish now with publishNow: true; accepts mediaUrls (fast media path), platformSettings and first comments |
get_post |
Get post group details |
update_post |
Edit a draft/scheduled post: replace content or the platforms target set, change status/schedule, publish now with publishNow: true, append mediaUrls, or merge platformSettings |
delete_post |
Delete a post group and all platform posts |
Media
| Tool | Description |
|---|---|
attach_media |
Attach a file from ChatGPT (for example, one the user uploaded) with an optional SEO fileName; Publora downloads and checks it and leaves the post as a draft |
get_upload_url |
Get a presigned S3 URL for media upload |
complete_media |
Finalize a file uploaded via get_upload_url |
delete_media |
Remove a media slot from a post |
prune_media_reference |
Remove a stale media reference that delete_media cannot handle (fixes MEDIA_REFERENCE_MISSING) |
Account and Connections
| Tool | Description |
|---|---|
account_context |
Your plan, available features, publishing quotas and schedule horizon — check before a batch of posts or after a quota error |
list_connections |
List all connected social media accounts |
Statistics (Mastodon and Bluesky)
| Tool | Description |
|---|---|
post_stats |
Engagement counters for published Mastodon or Bluesky posts |
profile_stats |
Followers, following and post count of a connected Mastodon or Bluesky account |
LinkedIn Engagement
| Tool | Description |
|---|---|
linkedin_create_reaction |
React to a post (LIKE, PRAISE, EMPATHY, etc.) |
linkedin_delete_reaction |
Remove a reaction |
linkedin_create_comment |
Post a comment on a LinkedIn post |
linkedin_delete_comment |
Delete a comment from a LinkedIn post |
linkedin_create_reshare |
Reshare a LinkedIn post, optionally with commentary and visibility settings |
linkedin_list_mentionables |
List mentionable LinkedIn people with ready-to-paste mention tokens (paid plans) |
The two statistics tools need a plan with analytics (Pro, or a legacy Premium plan). LinkedIn analytics/followers/profile-summary and the managed-user workspace endpoints (
/api/v1/workspace/*) are not MCP tools — use the REST OpenAPI reference. LinkedIn feed-retrieval tools (linkedin_posts,linkedin_post_comments,linkedin_post_reactions) are implemented but disabled pending LinkedIn'sr_member_socialapproval.
Example Conversations
Note: Analytics examples below illustrate REST-API capabilities. Over MCP, statistics reads cover only Mastodon and Bluesky (
post_stats,profile_stats); LinkedIn metrics come from the REST OpenAPI reference.
Schedule a Post
You: Schedule a post to my LinkedIn saying "Excited to announce our new product launch!" for tomorrow at 9am EST
Claude: I'll schedule that post for you. Let me first check your connected accounts...
Done! Post scheduled to LinkedIn for 2026-02-20 at 9am EST.
Cross-Platform Posting
You: Post "We're hiring! Check out our careers page" to Twitter, LinkedIn, and Bluesky right now
Claude: I'll post that to all three platforms...
Done! Published to:
- Twitter: https://twitter.com/...
- LinkedIn: https://linkedin.com/...
- Bluesky: https://bsky.app/...
Check Post Analytics
You: How did my last LinkedIn post perform?
Claude: Your post from Feb 15 received:
- 1,250 impressions
- 680 unique viewers
- 28 reactions, 5 comments, 3 reshares
- Engagement rate: 2.9%
Content Planning
You: Show me all my scheduled posts for next week
Claude: You have 5 scheduled posts:
- Mon 9am - LinkedIn: "Product update..."
- Tue 2pm - Twitter + LinkedIn: "Blog post..."
- Wed 10am - LinkedIn: "Case study..."
- Thu 3pm - Twitter: "Quick tip..."
- Fri 9am - LinkedIn + Bluesky: "Week recap..."
Campaign Management
You: Cancel all my draft posts and show me what's still scheduled
Claude: Deleted 3 draft posts. You still have 2 scheduled:
- Tomorrow 10am - LinkedIn: "Q1 results..."
- Friday 2pm - Twitter: "Weekend tips..."
Weekly Review
You: Give me a summary of how my LinkedIn performed this week
Claude: This week's LinkedIn performance:
- 5 posts published
- 4,230 total impressions
- 89 reactions, 12 comments
- Best performer: "Product launch" (1,850 impressions)
- Follower growth: +47
Python Client Example
For programmatic access or testing:
import asyncio
from mcp import ClientSession
from mcp.client.streamable_http import streamablehttp_client
async def main():
headers = {"Authorization": "Bearer sk_YOUR_API_KEY"}
async with streamablehttp_client("https://mcp.publora.com", headers=headers) as (read, write, _):
async with ClientSession(read, write) as session:
await session.initialize()
# List tools
tools = await session.list_tools()
print(f"Available: {len(tools.tools)} tools")
# Get connections
result = await session.call_tool("list_connections", {})
print(result.content[0].text)
asyncio.run(main())Install dependencies:
pip install mcp httpxVerification
Health Check
curl https://mcp.publora.com/health
# {"status":"ok","service":"publora-mcp"}Test MCP Handshake
curl -X POST https://mcp.publora.com \
-H "Content-Type: application/json" \
-H "Accept: application/json, text/event-stream" \
-H "Authorization: Bearer sk_YOUR_API_KEY" \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"test","version":"1.0.0"}}}'Authentication
Tool calls are authenticated in one of two ways:
- OAuth 2.1 sign-in — Claude, Claude Code and other browser-capable clients open a Publora window; approving it mints a dedicated key for that client.
- Your Publora API key (the same key as for the REST API) via
Authorization: Bearer sk_...(recommended) orx-publora-key: sk_....
HTTP Endpoints
| Method | Path | Purpose |
|---|---|---|
| POST | /mcp or / |
MCP messages (Streamable HTTP) |
| GET | /mcp or / |
SSE stream (server notifications) |
| DELETE | /mcp or / |
Close session |
| GET | /health |
Health check |
Troubleshooting
"API key required" Error
Ensure your Authorization header is correctly formatted:
- Correct:
Authorization: Bearer sk_abc123... - Wrong:
Authorization: sk_abc123...
Tools Not Showing
- Check your MCP config syntax (valid JSON)
- Verify the
typeis"http"(not"url") - Restart your MCP client completely
- Run
/mcpin Claude Code to verify connection
Session Errors
MCP uses sessions for stateful connections. If you see "Invalid or missing session", the server will automatically create a new one on the next request.
Comparison: MCP vs REST API
| Feature | MCP Server | REST API |
|---|---|---|
| Interface | Natural language via AI | HTTP requests |
| Best for | Interactive exploration, quick tasks | Programmatic integration, automation |
| Setup | One-time config | Per-request auth |
| Tools | 20 personal tools + 12 Agency tools (32 total; Agency account required for Agency tools) | Full API access |
| Rate limits | Same as REST API | Same |
Use MCP for conversational workflows. Use REST API for production integrations.
Publora is built by Creative Content Crafts, Inc. Need AI-powered content creation for LinkedIn, Threads, and X? Try Co.Actor — the best AI service for authentic thought leadership at scale.