MCP server

Schedule your posts from any AI assistant

ProPost speaks the Model Context Protocol (MCP), the open standard AI assistants use to work with other apps. Connect Claude, ChatGPT, Cursor, VS Code, Gemini or any MCP client, and ask it to draft, schedule and publish on all your channels. Your plan, its limits and your approval rules apply to everything it does.

Server URL

https://www.propost-ai.com/api/mcp

Before you start

A ProPost Pro or Premium plan. Every request checks that the plan is active; an expired plan stops the connection until it's renewed.
At least one connected channel (Settings → Channels).
An assistant that supports remote MCP servers. Most sign you in with OAuth; for the rest, create an API key in Settings → Integrations.

You choose how far it may go

When you connect an assistant (or create a key) you pick one level. ProPost enforces it on every request, whatever the assistant tries.

Drafts only

Writes and edits drafts. You review and schedule them in ProPost.

Schedule

Also schedules and queues posts. You can still cancel before they go out.

Full autopilot

Also publishes immediately. Posts go public with no review.

Change it or disconnect at any time in Settings → Integrations.

In Claude and other assistants that honour MCP's safety hints, you're asked to confirm before a post is edited or deleted, or a posting schedule is replaced. Writing drafts, scheduling, queueing, unscheduling and publishing run without a prompt, within the level you chose: only Full autopilot can publish right away.

Set up your assistant

Pick your assistant. Where it says sign in, you'll be sent to ProPost to approve the connection and choose its level.

claude.ai, Claude Desktop and mobile

Steps

Open Settings → Connectors and choose Add custom connector.
Name it ProPost and paste the server URL.
Click Connect, sign in to ProPost and choose what Claude may do.
In a chat, turn on the ProPost connector from the tools menu and ask, for example: "Draft a LinkedIn post about our launch and queue it."

Server URL

https://www.propost-ai.com/api/mcp

On Team and Enterprise plans an owner may need to add the connector for the organization first.

Tools

What the assistant can call. Read tools need nothing more than a connection; write tools also respect the level you chose.

get_account

Needs: Any level

Your plan, renewal date, AI drafts used and left this month, weekly scheduling limit, channel counts, time zone, language, and this connection's level. Assistants call it first.

Counts against: Hourly request limit

list_channels

Needs: Any level

Your connected channels with their id, network, name, character limit, whether they need an image or video, and whether they need reconnecting.

Counts against: Hourly request limit

list_posts

Needs: Any level

Your posts, filtered by status (draft, approved, scheduled, published, failed) and channel, up to 100 at a time, each with its engagement numbers when the network has reported them.

Counts against: Hourly request limit

get_post

Needs: Any level

One post with its status, text, hashtags, schedule time, live link or error.

Counts against: Hourly request limit

get_posting_schedule

Needs: Any level

A channel's weekly queue times, whether the queue is paused, and the next free slot.

Counts against: Hourly request limit

get_channel_stats

Needs: Any level

A channel's followers and totals (impressions, reach, likes, comments, shares, engagement rate) for the last 7, 30 or 90 days, compared with the period before, plus how the posts you published from ProPost did.

Counts against: Hourly request limit

list_personas

Needs: Any level

Your voices: the default one and each named persona, with the channels that write in it.

Counts against: Hourly request limit

check_post

Needs: Any level

Checks whether a post is ready to go out on its channel (length, required media, reconnection, plan allowance) without changing anything.

Counts against: Hourly request limit

suggest_posting_times

Needs: Any level

mcp.page.tools.suggest_posting_times.summary

Counts against: mcp.page.tools.suggest_posting_times.uses

list_ideas

Needs: Any level

Your idea bank, newest first.

Counts against: Hourly request limit

add_idea

Needs: Any level (write)

Saves an idea to your idea bank for later.

Counts against: Idea bank size on your plan

delete_idea

Needs: Any level (write)

Removes an idea from your idea bank.

Counts against: Hourly request limit

create_persona

Needs: Any level (write)

Defines a new voice: name, role, tone, audience, topics and up to three writing samples. Up to 10 per account.

Counts against: Hourly request limit

update_persona

Needs: Any level (write)

Changes a voice; only the fields given change. Channels using it write with the new settings from then on.

Counts against: Hourly request limit

set_channel_persona

Needs: Any level (write)

Sets which voice a channel's drafts are written in, or puts it back on the default voice.

Counts against: Hourly request limit

set_hashtags

Needs: Any level (write)

Writes hashtags at the end of a post's text so they're published with it, replacing an existing hashtag line rather than stacking.

Counts against: Channel character limit

attach_image

Needs: Any level (write)

Attaches an image to a post: a graphic the assistant designs as SVG (rendered by ProPost, with fonts bundled), an image sent directly (up to about 3 MB), or one fetched from a public https link (up to 8 MB). Stored like an upload in the app and published with the post.

Counts against: 8 MB per image

remove_image

Needs: Any level (write)

Removes the image from a post that isn't published yet.

Counts against: Hourly request limit

rewrite_post

Needs: Any level (write)

ProPost AI rewrites a post following an instruction, within the channel's limit. Up to 3 rewrites per post.

Counts against: Hourly AI limit (no monthly draft)

generate_hashtags

Needs: Any level (write)

Suggests hashtags for a post and saves them as suggestions; the text is unchanged.

Counts against: Hourly AI limit (no monthly draft)

copy_post_to_channel

Needs: Any level (write)

Copies a post onto another of your channels, condensed to that network's limit if needed.

Counts against: Hourly AI limit when condensing

create_draft

Needs: Any level (write)

Saves a post the assistant wrote, on one or several channels (a linked copy per channel, each within that network's limit). On Drafts only, it waits for your approval.

Counts against: Hourly request limit

generate_draft

Needs: Any level (write)

ProPost's own writer drafts a post from an idea, in your voice and persona, with copies condensed for each channel.

Counts against: 1 monthly AI draft, hourly AI limit

update_post

Needs: Any level (write)

Changes a post's text or hashtags (not once published). On Drafts only, an edited post goes back to draft for your review.

Counts against: Hourly request limit

unschedule_post

Needs: Any level (write)

Takes a post off the schedule so it isn't published.

Counts against: Hourly request limit

delete_post

Needs: Any level (write)

Deletes a post that isn't published yet.

Counts against: Hourly request limit

schedule_post

Needs: Schedule or Full autopilot

Schedules a post for an exact time.

Counts against: Weekly scheduling limit, channel allowance

add_to_queue

Needs: Schedule or Full autopilot

Puts a post in the channel's next free queue slot.

Counts against: Weekly scheduling limit, channel allowance

set_posting_schedule

Needs: Schedule or Full autopilot

Replaces a channel's weekly queue times, or pauses and resumes its queue.

Counts against: Hourly request limit

publish_now

Needs: Full autopilot

Publishes a post right away. Only on connections set to Full autopilot; it runs without a confirmation prompt there.

Counts against: Channel allowance

Your plan's limits still apply

Plan checked on every request: Pro or Premium, active and not expired.
AI drafts: generate_draft uses your monthly allowance (300 on Pro, 1,500 on Premium) and the hourly AI limit, exactly like Generate in the app.
Requests: up to 300 tool calls an hour on Pro and 1,000 on Premium.
Posts: each network's character limit, required images or videos, your channel allowance, and channels that need reconnecting are all checked before anything is scheduled or sent.
Errors come back with a clear code (for example quota_exceeded or autonomy_limit) and what to do next, so the assistant can tell you.

Security

An assistant only ever reaches the ProPost account that approved it. It can't see or change anyone else's posts, channels or data; anything from another account is reported as not found.
Your contacts, inbox messages, comments, payment details and connected-account credentials aren't available through MCP at all.
Sign-in is OAuth 2.1 with PKCE. ProPost never sees your assistant's password and the assistant never sees yours.
Tokens are stored only as a hash, expire after an hour and refresh automatically; a reused refresh token disconnects the app.
API keys are shown once. Disconnect an app or delete a key in Settings → Integrations and it stops working immediately.
Everything an assistant does is recorded in your account's activity log with its name.

Questions

Can the assistant post without me?

Only if you chose Full autopilot. On Drafts only it can't schedule or publish anything; on Schedule it can schedule but not publish immediately, and you can cancel before the time.

Why did it say quota_exceeded?

Your monthly AI drafts are used up. The assistant can still write the post itself and save it with create_draft, which uses no allowance.

It says plan_required.

The account is on Free or the paid plan has expired. Renew or upgrade in Settings → Plans & billing; the connection works again straight away.

What times does it use?

Your ProPost time zone (Settings → Preferences). Assistants read it with get_account and send times with an offset, so "tomorrow 9am" means 9am where you are.

Ready to connect?

Open Settings → Integrations for your server URL and keys.