Spotlight: Letmepost — the social API that pre-eats platform reviews
If you have ever tried to ship a "post to social media" feature, you know the actual work isn't the posting. It's the OAuth dance for each platform, the branching logic for what each API accepts, the weeks-long wait for Meta App Review or LinkedIn's Marketing Developer Platform to approve your app, and the vague 500 errors you get when you finally go live. Letmepost is built specifically around that pain, not around the posting itself.
The pitch is simple: one POST request, fanned out to eight platforms (X, Bluesky, and Pinterest are live now, the rest are sitting in platform review), with a structured result back per destination. No per-platform SDKs, no guessing which field a given API wants this week.
The smart move here isn't the fan-out, plenty of tools do that. It's that letmepost positions itself as "the reviewed app of record." Instead of every developer independently suffering through Meta's and LinkedIn's review queues, you connect through letmepost's already-approved OAuth apps. That's a real wedge: it turns a shared, annoying, one-time cost (getting reviewed) into something one company eats once and resells to everyone else. It's also why the roadmap is honest about what's live versus what's pending review instead of pretending everything works today.
The other detail worth noting is the failure model. Eighty preflight rules run locally before anything hits a platform, so a bad request gets rejected with the exact rule that fired and a docs link, rather than an opaque 500. Every write also accepts an Idempotency-Key, with retries returning the cached response for 24 hours, so a flaky network connection doesn't turn into a duplicate post. These are the kind of unglamorous guarantees you only appreciate after you've been burned by their absence, which suggests whoever built this has been burned before.
It's Apache 2.0 from the first commit, and the hosted version is the same image you can self-host. Pricing is flat per organization ($0 self-host tier, then $79/mo and $299/mo hosted tiers), not per seat or per profile, and the only metered thing is posts published, with warnings at 80% and 100% of your cap. Combined with a native MCP server (a hosted endpoint plus a stdio binary on npm, 21 tools autogenerated from their OpenAPI spec) that lets Claude, Cursor, and other MCP-aware agents drive the whole thing, it's clearly aimed as much at AI agent builders as at traditional app developers.
Who should try this: developers or small teams building a product feature that needs to publish to multiple social platforms, especially if you don't want to own the OAuth and review-queue overhead yourselves. It's also a genuinely interesting pick for anyone building agents that need to post on a user's behalf, given the MCP-native design. If you're self-hosting and comfortable with open source infrastructure, the free tier removes most of the risk of trying it.
Who should skip it: if you only need to post to one or two platforms you already have deep API access to, this is more machinery than you need. And if your use case depends on a platform that's still "in review" (Meta and LinkedIn aren't live yet), you'll be waiting, same as everyone else, just with less of your own effort spent on it.
Where this goes next, in my read: the value of "the reviewed app of record" only compounds as more platforms clear review, so the real test is execution speed on that queue, not the API design, which already looks solid. If Facebook, Instagram, and LinkedIn land in the next few months, this becomes a much easier default recommendation. If they stall, the pitch weakens fast, since developers can route around one or two platforms themselves.
Try Letmepost: letmepost.dev
See the launch: Letmepost on welaunch.sh