Schedule single posts and threads to X from one calendar, or from your agent. This guide covers what planpo.st sends to X, the thread craft, timing, and what X does not allow.
Free plan, no card: 2 accounts, 30 posts a month, your agent included. X needs a paid plan.
What planpo.st publishes to X (Twitter): Text posts and threads.
X is the network where builders, founders and the people who buy from them actually talk. It is also the network most hostile to inconsistency: the feed moves fast, memory is short, and an account that posts in bursts then disappears earns exactly the reach it planned for.
The difference between the accounts that grow and the ones that stall is rarely talent. It is showing up daily, which is a scheduling problem. This guide covers running X through planpo.st end to end: single posts, proper threads, timing, what to write when you have nothing, and how to hand the daily writing to an agent.
For software, indie products and B2B services, X compresses the distance between you and the exact people who buy: founders follow founders, developers follow developers, and a good post travels through that graph in hours. It is the only network where build-in-public is a genre with an audience waiting for it, and where the reply section routinely contains your next customer, hire or partner.
The cost is volume sensitivity. X rewards frequency more than any network here: one post a week disappears, one a day compounds. That cadence is unsustainable live and completely sustainable scheduled, which is why X accounts benefit from a queue more than any other channel planpo.st supports.
Who this channel is for
One idea inside 280 characters. The constraint is the craft: a specific observation, a number, a question. Single posts are the daily heartbeat of an account, cheap to write and batchable ten at a time.
A chain of connected posts published in order, planpo.st handles the chaining. Threads are X's long-form: how you built something, what a failure taught you, a process in steps. One strong thread earns more followers than a month of singles, because threads get bookmarked and shared as units.
Images and video, within X's rules
Thread structure that consistently works: the first post is a complete, specific promise (a number helps), each middle post carries exactly one point and can survive being screenshot alone, and the last post lands the conclusion plus a plain ask (a reply, a name to search). planpo.st counts characters per segment, so an over-long segment gets caught while you write, not when the chain publishes.
Open Connections and pick X
planpo.st uses X's OAuth 2.0 flow with PKCE, the current official method. You approve on X's own page; no password ever touches planpo.st. X needs a paid plan: Creator or Pro, or a trial of either.
Approve the requested scopes
Posting and reading your own account. Tokens are stored encrypted and refreshed automatically in the background, so the connection does not rot between posts.
Send a test post
Schedule one post two minutes out. Thirty seconds of paranoia now saves a missed launch post later.
Several X accounts work side by side, for example a product account and a founder account, each picked per post.
Daily posting does not require daily writing. The pattern that works: one session per week where you write ten to fifteen singles and one thread, then spread them across the calendar at your best times. The composer's per-segment counter makes threads fast to draft, and moving a post to another slot is a quick edit when something newsworthy lands mid-week.
For a launch, work backwards: the announcement thread on launch day, teaser singles the two days before, results posts the week after, all scheduled in one sitting while the energy is high. The queue holds the plan so launch morning is about talking to people in the replies, not typing the announcement with shaking hands.
Replies are the live part
X's half-life is the shortest of any network here: a post does most of its lifetime numbers within hours. That makes timing matter more, and it makes your audience's timezone the variable that dominates. A US developer audience and a European founder audience are seven hours apart, and a generic best-times chart averages them into uselessness.
planpo.st's best-times suggestion reads your own engagement history. The bootstrap heuristic until you have data: weekday mornings in your audience's timezone for reach, and thread launches early on weekend mornings, when feeds are slower and long reads get their chance. Then let your numbers overrule the heuristic, because they will.
The weekly ship log: what you built, with one real number
Build-in-public's bread and butter; an agent can draft it from your commits.
A thread on the worst mistake you made at your current stage
Failure content outperforms success content and is infinitely honest.
Before-and-after metrics of one change (with the change named)
Specific numbers get bookmarked; vague lessons get scrolled.
One contrarian opinion about your industry, argued in one post
Gentle disagreement is X's engagement engine. Pick fights with ideas, not people.
A tool or workflow you use daily, in one post with the why
Utility posts travel through the follower graph farther than opinions.
The pricing thread: what you charge, what you tried, what happened
Money transparency is rare enough to be remarkable.
Screenshot of a customer message that made your week (with permission)
Social proof with a heartbeat.
A 'how it started / how it's going' pair with real dates
Progress narrative in two lines; schedule it for milestones.
One question you genuinely want answered by your audience
Questions earn replies, replies earn reach, and you learn something.
The teardown thread: how a product you admire does one thing well
Borrow an established name's gravity while showing your taste.
Your most-asked customer question, answered in one post
You already wrote this answer five times in support. Reuse it.
A monthly revenue or usage recap, honest numbers, no rounding up
The honesty is the hook, and the monthly rhythm schedules itself.
X pairs naturally with Threads and LinkedIn, and planpo.st turns one draft into all three: the X version tightened to 280, Threads at 500 with a more conversational register, LinkedIn expanded toward its 3,000-character room. Write the idea once, let each network get its native cut, and edit any version before it goes.
A thread has a second life as a LinkedIn post (merge the segments into paragraphs) and a third as a blog post. Rewriting for another network is a job an agent does well: ask it to turn the thread into a LinkedIn post and it drafts the new version, for your approval if you want it. A good X thread is never one post; it is a content week.
For builders this is the natural fit: the agent that watches your repo can also narrate it.
You say
Look at this week's merged PRs, pick the three most interesting, and schedule a post about each for next week, morning slots. Keep my usual tone, no hashtags.
What happens
Claude Code or Cursor reads the actual changes, drafts three posts inside 280 characters each, and books morning slots via the best-times data. The posts are technically accurate because they came from the diff, not from memory.
You say
Draft a 6-post thread from my blog post at this URL and schedule it Saturday 9am.
What happens
The agent restructures the article into a thread with a strong opener and one point per segment, leaves the URL out of the X copy, chains it as a single thread post, and schedules it. You edit any segment before Saturday, and with approval on you approve it first.
To set this up, connect your agent once: one command for Claude Code, a link for Claude, ChatGPT and Cursor, and no key to paste. Then posting to X from Claude Code lists the settings the publisher reads and the limits that stop a post.
Per post, planpo.st cannot tell you, and here is why. X's API bills a post that contains a link at many times the price of one without, and every thread segment and reply is billed the same way. So planpo.st sends no links to X at all: a link in the X version stops the post when you schedule it. Without a link there is no tracked click for a payment to match.
What still works
For build-in-public accounts that is an honest split of jobs: X carries the conversation and the daily presence, and the networks that allow a tracked link carry the measurable ask.
Three posts on a motivated Monday, nothing for two weeks. The feed forgets you between bursts. A modest scheduled daily beat wins.
Segment one is the whole sale. If it does not carry a specific promise or number, the other segments never get read.
An account that only asks gets muted. The working ratio is several value posts for each ask.
Reach without relevance grows an audience that never buys, and a growing follower count hides it.
Applause from peers feels like traction. Check whether the accounts replying could ever be customers.
X is a volume game with a long tail. The flop cost nothing; the missed week after it cost the compounding.
Any agent that speaks MCP can, through the planpo.st server. These 8 have a page each for X, with the connect step, the settings the publisher reads and the limits that stop a post.
Free plan, no card: 2 accounts, 30 posts a month, your agent included. X needs a paid plan.