Bulk scheduling stops being a UI problem the moment your queue comes from somewhere else — a content database, a release pipeline, a client sheet. PostMCP AI gives you the API to generate it, the pre-flight to rehearse it, and the per-profile tracking to repair it. Same seventeen tools in the dashboard, in your script, and in Claude.
Free plan, no credit card — 5 profiles and 20 credits a month.
# generate a week of posts from your own data
for row in schedule.csv:
POST /api/tools/preflight_post # rehearse
POST /api/tools/create_post # commit
# or collapse it into batches of 20
POST /api/tools/multicall
{ "calls": [ … ], "stopOnError": true }
→ { "ok": 18, "failed": 1, "skipped": 1 }CSV import is fine until you want the queue generated from something — a content database, a release pipeline, a client spreadsheet that changes weekly. multicall runs up to 20 operations per request, validates every tool name before executing, and returns one labelled result per call so you can reconcile a batch programmatically.
When you push 200 posts, some will fail — a token expired, an image 404’d, a network throttled you. Every delivery is tracked per profile, and re-publishing skips the ones that already landed. A failed batch is repaired, not re-run.
A ten-target broadcast becomes ten posts with ten IDs. You can edit, reschedule, retry or cancel any one of them without touching the other nine — which is what makes volume manageable rather than merely possible.
A dry run reports character overruns, unconnected profiles, missing media on networks that require it and the exact credit cost. Running it over a batch before committing is the difference between a 200-post publish and a 200-post incident.
| Plan | Monthly | Credits | Profiles | Workspaces |
|---|---|---|---|---|
| Free | $0 | 20 | 5 | 1 |
| Starter | $20 · $16 annual | 2,000 | 10 | 3 |
| Professional | $50 · $40 annual | 6,000 | 25 | 5 |
| Business | $100 · $80 annual | 14,000 | 50 | 10 |
1 credit per profile delivered to, 5 for X, plus a one-off 50-credit surcharge on posts containing a link. One-time packs top up without changing your plan. Full pricing.
Figures here describe PostMCP AI. Other tools change plans often — compare against their current pricing page rather than a third-party summary.
Most schedulers bolt an API onto a finished dashboard. PostMCP ships the same seventeen tools three ways: the dashboard calls them, your script calls them over plain HTTP, and an AI assistant calls them over MCP. Drafting, pre-flight, scheduling and retries are the same operations wherever they are invoked — with an approval step before anything reaches a live account.
npx @postmcpai/server
Open source under MIT. Runs over stdio or streamable HTTP.
Short, because most of this decision comes down to one question: do you need a finished dashboard, or something your own code can drive? These land on different sides of it.
Every tool the dashboard uses is a REST endpoint and an MCP tool, so a script, a CI job or Claude can publish. Free plan, no card.
The lightweight default. Small free tier, no meaningful API for building on.
Category-based queues that keep re-posting evergreen content on a rotation.
Broad network coverage with reporting, aimed at agencies watching the invoice.
Strong reporting and competitor tracking with a usable free tier.
Built around reusable content libraries that refill the queue automatically.
Developer-first with no real dashboard. The cleaner answer if you only ever want an endpoint.
Positioning notes, not a feature audit. Every product here — ours included — changes plans and limits faster than a page like this can track.
Five connected profiles, 20 credits a month and the full calendar, with no credit card. The API and the MCP server are not a paid upgrade — you get a key on the free plan, which is the only honest way to let someone find out whether a publishing API fits their pipeline.