Let Claude, Cursor, or any MCP client draft, schedule, and publish across seven networks, then ask it how much revenue those posts brought in. Attribution lives behind the same API the agent calls, so the loop closes in one conversation.
45 tools · 7 networks · approval guardrail on by default
› Post the new feature announcement to LinkedIn and X tomorrow 9am.
Scheduled 2 posts · Tue 09:00 CET
Tracked link: planpo.st/r/f3a
› How did last month's posts do?
€4,182 attributed · 214 conversions
Best: LinkedIn carousel, 12 Jul: €1,340
Publishing is the easy half. Knowing what the publishing was worth is the half everyone else leaves out.
Ask in plain language. Your agent drafts per platform, respects your brand voice, and schedules into the calendar you already use.
Eight networks with native formats: threads, reels, shorts and stories. It renders video and carousels too, not just text.
Every post carries a tracked link. Ask what last month earned and the answer comes back in the same conversation.
Create an API key in Settings, then paste one of these. The hosted endpoint needs nothing installed.
claude mcp add --transport http planpost https://mcp.planpo.st/mcp --header "Authorization: Bearer pk_live_xxx"Nothing to install. Your agent talks to our endpoint and your key is sent with each request.
Replace pk_live_xxx with a key you created. Existing keys are stored hashed, so we can't show them again.
On by default. Agent posts wait in your Approvals inbox until you say go.
A read-only key cannot publish. Grant exactly the permissions the agent needs.
The MCP server is MIT-licensed on npm as planpost-mcp. Read it before you run it.
› Draft this week's posts from my latest blog article and schedule them for 9am.
Pulled the article, wrote six native drafts, scheduled across LinkedIn, X and Instagram.
› Turn that testimonial into a quote video and publish it as a reel.
Rendered 9:16 from a template, queued it, returned the render status.
› Which posts made money last month?
Returned revenue by post and platform, ranked, with conversion counts.
Most agent integrations stop at “create a post”. Yours can render a video, build a branded carousel, route a draft through approval, and read what any of it earned.
list_posts · get_post · create_post · update_post · delete_post · preview_post · publish_post · bulk_create_posts · delete_posts · list_campaigns
list_connections · list_account_groups · get_account_group · create_account_group · update_account_group · delete_account_group · add_connections_to_group · remove_connections_from_group · get_connection_link
create_video · get_video_status · list_video_templates · generate_carousel · render_carousel · get_carousel_render · list_carousel_styles · list_brand_kits · generate_image
get_revenue_overview · get_analytics · get_best_times · suggest_schedule
submit_for_approval · list_pending_approvals · get_approval_status · get_brand_voices · get_brand_voice_prompt
list_content_sources · add_content_source · delete_content_source · fetch_source_content · list_imported_content · repurpose_content · upload_media · list_media
It is a scheduler an AI agent can operate directly, rather than one you click through yourself. Planpo.st exposes an MCP server and a public API, so Claude, Cursor, or any MCP-capable client can draft, schedule, publish, and read analytics on your behalf.
Any client that speaks the Model Context Protocol. That includes Claude Code, Claude Desktop, and Cursor. The hosted endpoint needs no local install; the npm package runs over stdio if you prefer the key to stay on your machine.
Only if you let it. Every API key has an approval guardrail, on by default: posts the agent creates land in your Approvals inbox and go live once you approve them. Turn it off per key when you trust the automation.
Revenue attribution sits behind the same API the agent calls. Most tools can tell an agent that a post published; Planpo.st can tell it what the post earned, so the agent can act on results instead of just impressions.
No. Create an API key in Settings, paste one connect command into your agent, and talk to it in plain language.
Seven days free. Connect an agent in a minute, and find out which posts pay for themselves.
Per-agent guides, with real prompts
The rest of the loop