The LinkedIn MCP server
Schedule & Chill ships a native Model Context Protocol server at https://schedulenchill.com/mcp-oauth. Paste that URL into Claude, ChatGPT, Cursor or any MCP client, approve the consent screen, and your agent gains 22 tools to schedule, edit, cancel, and inspect posts across LinkedIn, Bluesky, Mastodon, YouTube and TikTok — no API key, no glue code.
Free · No credit card · No trial timer
- Endpoints
https://schedulenchill.com/mcp-oauthOAuth — no keyhttps://schedulenchill.com/mcpAPI key- Auth
- OAuth 2.1 with PKCE and dynamic client registration, so a connector configures itself from the URL.
- Platforms
- LinkedIn, Bluesky, Mastodon, YouTube and TikTok.
22
Tools
4
Resources
2
Prompts
Connect with one URL
No API key, nothing to paste into a config file. Add a connector in Claude, ChatGPT, Cursor, VS Code or Zapier and give it this address.
https://schedulenchill.com/mcp-oauth- Claude
- ChatGPT
- Cursor
- VS Code
- Zapier
Your tool discovers the rest on its own, sends you to a consent screen naming the tool and what it is asking for, and gets its own token once you approve. Each tool is approved separately and can be revoked on its own.
For clients that want a static header, and for scripts. Generate a key under Settings → API Keys and drop this into your MCP client.
{
"mcpServers": {
"schedulenchill": {
"url": "https://schedulenchill.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Either way, ask your agent “what social accounts do I have connected?” — that calls get_accounts and confirms the link works.
One sentence in, a scheduled post out
The real tool names and fields your agent sees. Hover to pause, or pick a scenario.
Schedule this for Tuesday 9am on LinkedIn: “We just shipped per-account delivery status.”
{}result
{
"count": 2,
"accounts": [
{ "id": 789, "platform": "linkedin", "health": "ok" },
{ "id": 790, "platform": "bluesky", "health": "ok" }
]
}{
"content": "We just shipped per-account delivery status.",
"account_ids": [789],
"when": "2026-10-27T09:00:00-04:00"
}result
{
"post_id": 1042,
"status": "scheduled",
"scheduled_at_local": "2026-10-27T09:00:00-04:00",
"timezone": "America/New_York"
}Done. It goes out on LinkedIn Tuesday at 9:00 AM. You can cancel it any time before then.
22 tools your agent gets
Every tool is typed, documented, and maintained alongside the REST API. Grouped by what you would actually ask for.
Create and publish
5Everything that puts words on a platform.
schedule_postCreate a post for one or many accounts, now or at a future time.update_postEdit a draft or scheduled post before it publishes.publish_nowSend a draft, scheduled or failed post immediately instead of waiting.retry_postResend only the channels that failed, leaving published ones untouched.duplicate_postCopy a post — content, channels and media — into a new draft.
Read and undo the queue
6What is planned, what happened, and how to take it back.
list_postsList posts by status — the queue, drafts, published or failed.list_scheduledRead the upcoming queue so the agent knows what is already planned.get_postRetrieve one post with per-account delivery status and errors.cancel_postCancel a scheduled post the agent or user decides against.delete_postRemove a post in any state. Soft delete, so nothing is lost.restore_postUndo a cancel or delete and bring the post back.
Accounts and platform rules
5What the agent is allowed to do, before it tries.
get_accountsList connected accounts with ids, platforms, and handles.get_groupsList channel groups so the agent can target a set of accounts at once.get_capabilitiesDiscover live platforms, their native option schemas, and media limits.get_queue_slotsRead the next open slots in the posting schedule to offer real times.get_tiktok_creator_infoRead TikTok posting limits and privacy options before scheduling.
Media
4An asset library the agent can search by meaning, not filename.
browse_mediaSearch the media library for an image or video to attach.upload_mediaIngest a file into the media library from a public URL.update_mediaWrite a description and tags back onto a file so it can be found later.list_asset_vocabularyList the tags and descriptions that exist, so a search uses real terms.
Insight
2Real numbers from the platforms, not just counts of what was sent.
get_statsRead publishing stats across accounts and time ranges.get_analyticsRead real engagement — reach, impressions, likes, top posts — by channel.
Context an agent can pull in
4 read-only resources
Context an agent can load without changing anything.
schedulenchill://accounts/connectedActive connected accounts.
schedulenchill://schedule/upcomingThe next 20 scheduled posts.
schedulenchill://media/libraryThe 20 most recent media files.
schedulenchill://get-startedOnboarding context for a fresh agent.
2 prompts
Ready-made templates an MCP client can offer the user.
content_ideaGenerate post ideas for a niche.
draft_postDraft a post about a topic, sized and toned for the target platforms.
A typical agent flow
- 1
Discover
get_accounts returns where the agent is allowed to post, with real ids.
- 2
Attach
browse_media finds an existing asset, or upload_media ingests one from a URL.
- 3
Schedule
schedule_post takes the content, target accounts, and an optional publish time.
- 4
Verify
list_scheduled confirms the queue; get_post shows per-account delivery status.
MCP and social posting
An MCP server is a standard way to expose tools to an AI agent. A social media MCP server gives the agent typed tools for posting — schedule, cancel, browse media, read the queue — so it can act on social platforms without you writing custom function definitions or platform OAuth code.
22 tools, 4 read-only resources, and 2 prompts. The tools cover scheduling, updating, inspecting, duplicating, publishing now, retrying failed channels, cancelling, deleting and restoring posts, listing connected accounts and channel groups, reading per-platform limits and queue slots, browsing, uploading and describing media, and reading both post counts and real engagement analytics.
Live for everyone today: LinkedIn, Bluesky, Mastodon, YouTube and TikTok. Adapters for X, LinkedIn Pages, Facebook, Instagram and Pinterest are built and tested but each is still in its platform's review queue, so the app shows them as unavailable rather than letting a post fail silently. Threads and Google Business Profile are on the roadmap.
Any client that supports remote MCP servers over HTTP. Claude, ChatGPT, Claude Desktop, Cursor, VS Code and Zapier connect over OAuth by pasting https://schedulenchill.com/mcp-oauth as a custom connector. Claude Code, Gemini CLI, Codex, Windsurf, Zed and n8n work with either that URL or an API key header. From your own code, use an MCP client library, or langchain-mcp-adapters if you are on LangChain.
Two ways, both giving access to the same tools. OAuth 2.1 with PKCE at https://schedulenchill.com/mcp-oauth is the recommended path: you paste the URL, approve a consent screen naming the tool, and it receives its own token — there is no key to create or store, each tool is revocable on its own, and your password is never shared. Or use the same API keys as the REST API at https://schedulenchill.com/mcp, sent as Authorization: Bearer YOUR_API_KEY.
It is built and maintained by Schedule & Chill, in the same codebase as the REST API. New API capabilities land in both surfaces, so the tools do not drift out of date.
No. Connecting a social account like LinkedIn or TikTok needs that platform’s own sign-in flow in a browser, which you complete once in the dashboard. That is separate from connecting the AI tool itself. After both, the agent can publish to any connected account.
Yes. Add https://schedulenchill.com/mcp-oauth under Settings → Connectors → Add and approve the consent screen. Adding a custom connector needs a paid ChatGPT plan with developer mode enabled. Earlier versions of this server were bearer-token only and could not be used from ChatGPT; that is no longer the case.
Have it schedule rather than publish. A future scheduled_at gives you a review window, and cancel_post removes anything you do not want. For programmatic control, add a human approval step before the publishing tool call.
No. The MCP server is free, like everything else. It shares your monthly post allowance with the REST API rather than metering separately.
Same capabilities, different consumers. The REST API is for deterministic code — workflows, cron jobs, your backend. The MCP server describes those same operations with schemas a model can reason about, so an agent can decide what to call.
Let it post, not just draft
Create a free account, connect a channel, and paste one URL into your AI tool. Two minutes, no API key.
Free · No credit card · No trial timer
Last updated September 4, 2026