How to schedule social media posts with Claude

Claude is good at writing in a consistent voice, which makes it a natural place to draft social posts. Once it is connected to a scheduling platform, it can also put those posts on the calendar, one version per network, without you copying anything.

To schedule social media posts with Claude, add a social media MCP server as a custom connector, sign in once, and ask Claude to draft and schedule. The platform holds each network's version to its rules and, with approval on, waits for a person before anything goes out. As of November 2026, custom connectors work on every Claude plan.

This guide uses Tracee as the example server. The Claude steps are the same for any remote MCP server.

Where can you use Claude to schedule posts?

You add the connector once on claude.ai or Claude Desktop, and it follows your account to the other apps. Anthropic's docs say remote connectors are brokered through your Claude account and the connection comes from Anthropic's cloud on every client (Claude help center, November 2026).

You also need a Tracee workspace with at least one connected social account. MCP access is included on every Tracee plan, from $10 a month at the time of writing (November 2026), with a 7-day trial that needs no credit card. See Tracee pricing.

How do you add Tracee to Claude?

On Free, Pro, Max or Ultra you add the connector to your own account. These steps follow Anthropic's guide to adding a connector that isn't in the directory, as of November 2026.

1. Open Customize > Connectors on claude.ai, or the same page in Claude Desktop.
2. Select Add, then Add custom connector.
3. Enter a name (Tracee) and the server URL https://traceeai.com/mcp, then click Continue.
4. Review the authentication settings. Claude reads the server's settings and marks them Detected; keep those.
5. Click Add.
6. Click Connect, sign in to Tracee in the window that opens, and allow access.
7. In a chat, click the + button, open Connectors, and make sure Tracee is on for that conversation.

About the OAuth client choice in step 4: Claude offers Use Claude's published identity, Register automatically (dynamic client registration) and Use your own OAuth client. Tracee supports dynamic client registration. If sign-in fails with the detected option, remove the connector and add it again with Register automatically. You can't change authentication settings on an existing connector; Anthropic's docs say to remove it and add it again.

On a Team or Enterprise plan

An Owner adds Tracee once for the organization:

1. Go to Organization settings > Connectors.
2. Select Add, then Custom. If asked for a type, choose Web.
3. Enter the name and https://traceeai.com/mcp, keep the detected sign-in settings, and click Add.

Members then open Customize > Connectors, find Tracee with the Custom label, and click Connect. Because each person signs in with their own Tracee account, Tracee's audit log shows which teammate's Claude did what. Avoid a shared credential here; Anthropic notes that with one, everyone gets that credential's access.

How do you write the first prompt?

Start with something that only reads, so you can see the connector work:

Using Tracee, which channels are connected, and what's scheduled for the next 7 days?

Then a real request with the facts Claude can't guess:

Using Tracee, read our brand rules, then write a post announcing our new reporting dashboard (https://example.com/launch) for LinkedIn and X. Audience: B2B marketing leads. Main point: they can now see which posts brought in contacts. Schedule it for Tuesday at 9:00 our time.

Claude will call brand to read your tone and banned phrases, channels to find the accounts, posts with action: draft to make one version per channel, and posts with action: schedule. The link goes out with UTM tags per channel, so contacts from it can be traced back to this post in your CRM.

Claude Projects help here. If you keep a project for social content, put your messaging notes and a few posts you liked in its knowledge, and start each request from that project. Tracee's brand rules still apply on top, because the server checks every draft.

How do you control what Claude may do?

Use two layers: Claude's tool permissions and the platform's guardrails. Claude asks before it runs a tool unless you chose Always allow, and you can set any tool to Blocked under Customize > Connectors by selecting the connector (Anthropic's security guidance).

Tracee groups actions into a few tools, so Claude's per-tool switch is coarse: the posts tool covers both drafting and scheduling. A sensible setup:

• Always allow read tools such as analytics, channels, brand and approvals.
• Ask each time for posts, media and inbox.
• Block apps and workflows if you only want Claude on social posts.

Fine-grained limits belong in Tracee, where Claude can't change them. In Settings > Agent rules a person decides whether agents may act at all, whether agent posts need approval (on by default), and which channels agents may post to. Deleting or taking down a post always waits for a person. Our guide to an approval workflow for AI agents has a policy template.

How do you review, approve and schedule?

Review every network's version. Tracee holds each one to that network's rules: the X version stays within 280 characters, Instagram needs an image or video, and TikTok and YouTube need a video. A version with a banned phrase is flagged and can't go out until fixed. Ask Claude "show me each version and any brand issues", then edit in plain words: "make the LinkedIn version open with the number".

When Claude asks to schedule, the answer is pending_approval, and Claude should tell you it is waiting:

1. A teammate opens the Approvals queue in Tracee and checks the post in the calendar.
2. Approve, and it goes out at the time Claude asked for, even if the chat is closed.
3. Reject, and nothing goes out. Claude sees the decision when you ask "did Tuesday's post get approved?"

Using Claude Desktop and mobile

Claude Desktop uses the same account connectors as the web, so there is nothing extra to set up. Desktop also supports local MCP servers through desktop extensions and a config file, but you don't need either for Tracee, because it is a remote server.

On iOS and Android, Tracee is available the next time you log in, once you have connected it on the web or desktop. Phones suit short checks: "what's waiting for approval?", "what went out today?", or "move Friday's post to Monday at 10:00".

Using Claude Code

Claude Code connects to the same server from your terminal. Per the Claude Code MCP docs:

claude mcp add --transport http tracee https://traceeai.com/mcp

Then type /mcp in Claude Code, pick tracee and choose Authenticate. Claude Code can now draft posts from files in your repository, such as a changelog. Developers who prefer scripts can have Claude Code run the Tracee CLI instead; we cover both in managing social media from Cursor and Claude Code.

Which prompts work well?

Troubleshooting

What are the limits?

Claude can only do what the platform's tools allow, and the platform can only do what each network's API allows. Tracee publishes to 13 networks, including LinkedIn, Instagram, X, TikTok, YouTube, Reddit and Google Business Profile. Each Tracee credential gets 120 requests a minute. And Claude writes what you ask for, so give it the link, the audience and the point. If MCP is new to you, our explainer on what a social media MCP server is covers the basics.

Next step

Add the connector, run one read-only prompt, then schedule one low-stakes post with approval on and follow it from Claude to the Approvals queue to the network. When that works, give Claude a week of source material and ask for the whole week in one request.