1---2name: content-calendar3description: >-4 Turns one playbook chapter, article or theme into a LinkedIn posting calendar at 3 posts a week. Reads the source5 through an operator lens and a buyer lens, pulls every topic that passes a strict quality bar plus the source's6 anchor framework, frames each with a pillar from the messaging house, a buyer role, a post shape and a hook line from7 company-profile.md, then stops for the reader to pick topics. After the pick it sequences them under balancing rules8 (strategic and tactical mix, shapes, formats, every buyer role each month, no duplicates) and ends with a9 tab-separated block that pastes into any sheet. Use it to plan founder or company posts from a source. Not for10 writing the posts (use founder-post) or repurposing a finished article into assets (use repurpose).11license: MIT12compatibility: >-13 Works in Claude, ChatGPT, Codex, Cursor and any agent that reads SKILL.md. Never pushes to a sheet or tool; the14 reader pastes the block. Python 3 is optional, for sequencing the calendar.15metadata:16 author: SHAKEUP17 version: "1.0.0"18 stage: make19 plugin: content20 homepage: https://www.shakeuphq.com/skills/content-calendar21---22 23# Content calendar24 25Finds every post worth writing in one source, lets the reader pick, and lays the picks out so no week repeats itself.26 27## Purpose28 29- One job: from one source, a picked set of framed topics in a balanced calendar.30- Stage: make. It ends in its own checks and a fix loop of at most 2 rounds, with one named stop for the pick.31- It never invents a topic, a figure or a reaction the source doesn't support, never builds the calendar before the reader picks, and never pushes anything to a sheet or a tool.32 33## Use when / not34 35| Situation | Use |36| --- | --- |37| A playbook chapter, guide section or article should become a month of posts | this skill |38| A theme already broken into topic ideas needs framing and a calendar | this skill, bare theme source |39| Picked topics need a calendar (the pick already happened) | this skill, from step 7 |40| Writing one of the posts | `founder-post`, from the calendar row |41| Only hooks | `hook-writer` |42| Posts, an email and scripts from one finished article, placed by channel | `repurpose` |43 44## Inputs45 46| Input | Required | Ways to provide, in order | If missing |47| --- | --- | --- | --- |48| The source: a chapter, an article or a bare theme | yes | a file or URL the agent can read; pasted text; an `article-write` or `repurpose` handoff | ask once; `BLOCKED` without it |49| The reader's pick | yes, for the calendar | the reader's answer at the checkpoint | stop at `AWAITING PICK` |50| Start date and posting days | no | the reader; `### Founder voice` > `Posting days` | next Monday; Tuesday, Wednesday, Thursday |51| The existing calendar | no | rows the reader pastes or exports from their sheet | duplicates checked within this run only; said so |52| Messaging house, buyers, moments | no | the profile | pillar `tactical` for all, personas from the source; coverage `partial` |53 54## Profile read55 56From `company-profile.md`: `## Product`, `## Buyers` and `### Buying group`, `## Buying moments`, `## Positioning`, `## Messaging house`, `## Voice` with `### Founder voice`, `## Claims and proof` with `### Never say`, `## Channels`, and every `## Learnings` entry for `content-calendar` and `founder-post`.57 58No profile: ask in one message for the buyer roles, the 2 or 3 messages the company stands for, and what it sells. Say that `company-profile` saves this next time.59 60## Steps61 621. **Apply learnings.** Picks the reader always skips (a `proposal` accepted as a rule) and archetype results (`fact`) shape the framing and the ranking.63 642. **Name the source type** (`references/extraction.md`): chapter, article or bare theme. One source per run.65 663. **Read through both lenses.** Fill the operator lens and the buyer lens fields, each grounded in a source line or a profile entry, or `none found`.67 684. **Extract.** Find the anchor first (and run its headline through the narrative test), then every nugget that passes all 7 points of the quality bar, at most 1 or 2 per named section of the source. Name the exact section each came from. If nothing passes, stop with `NO TOPICS`.69 705. **Frame every topic** (`references/angles-and-calendar.md`): core belief, persona, archetype, hook line, why it lands, what it means for the buyer, how we can sell it, motion, channel, format. Framing runs on every topic, not only the likely picks, so the reader picks with the full picture. A forced fit is written as `none found`.71 726. **Check the hook lines** before showing them: 18 words or fewer, no bait, no rhetorical question, no dash, a specific in each, and every figure traceable to the source or `## Claims and proof` (`claim-check`, or its test inline). Fix and recheck at most twice (Loop).73 747. **Checkpoint.** Show the topic preview (one small table per topic, anchor first) and ask which topics carry forward and whether the anchor is in. Stop: `AWAITING PICK`.75 768. **Sequence the picks.** Run:77 78 ```79 python3 scripts/calendar_balance.py --input picked.json80 ```81 82 Without Python, place topics slot by slot (3 a week) and check each placement against the balancing rules: no more than 2 strategic or 2 tactical in a row, no archetype twice in a week, no more than 1 anchor a week, no more than 2 of a format in a row, arcs back to back in order; then check every persona appears in each 4-week stretch and nothing repeats the existing calendar. Say which way you ran it. A rule the picks can't meet is reported, never forced.83 849. **Show the calendar** in the same per-topic table format, then the tab-separated block with every column on every row.85 86## Check and verdict87 88| Verdict | Condition (first match wins) |89| --- | --- |90| `BLOCKED` | No source |91| `NO TOPICS` | No nugget passes the quality bar: say what the source lacks (a consequence, a case, a figure, a position) |92| `AWAITING PICK` | Topics are framed and the reader hasn't picked yet |93| `READY WITH GAPS` | The calendar is built, but a balancing rule couldn't be met, a duplicate was flagged, a persona is missing from a 4-week stretch, or a hook still has an open item |94| `READY` | The calendar is built and every rule holds |95 96Named stops: `BLOCKED` asks for a source. `AWAITING PICK` is the checkpoint: it stops until the reader picks, on every run, including repeat runs. No other stops.97 98## Loop99 1001. Check the hook lines (step 6) and, after sequencing, the balancing rules (step 8).1012. Fix: rewrite a failed hook once, changing only what failed; re-sequence once if a rule failed and another order exists.1023. Recheck. At most 2 rounds; then hand over with what's still open, labelled. Never fix balance by inventing or splitting a topic.103 104## Learn105 106Append to `## Learnings` when:107 108- the reader's picks show a pattern across 2 or more runs, such as skipping every tactical topic (a `proposal`);109- posts from the calendar come back with results by archetype, pillar or persona (a `fact`, with the source);110- the reader changes the cadence or days (a `fact`, and update `### Founder voice` > `Posting days`).111 112Example: `- 2026-10-24 | content-calendar | proposal | Lead each month with a self-check list anchor | Why: anchors in that shape drew the most saves in 3 of 4 months (post analytics, Jun to Sep 2026) | Status: proposed`113 114## Output115 116At the checkpoint:117 118```119Verdict: AWAITING PICK · Coverage: <complete | partial (no messaging house; no buying moments)>120Source: <name, type> · Topics: <n> plus <n> anchor121 122**Anchor. <the framework, in one plain sentence>**123| Field | Value |124(pulled from, archetype, channel, core belief, motion, why it lands, hook line, what it means for the buyer, how we can sell it)125 126**1. <topic in one plain sentence>**127| Field | Value |128(repeat per topic)129 130Which topics carry forward (numbers or "all"), and is the anchor in? Start date and days, if not the defaults?131```132 133After the pick:134 135```136Verdict: <READY | READY WITH GAPS> · Sequenced by: <script | hand placement>137Cadence: <n> a week, <days>, from <date>138 139**Week 1, <day> <date>: <topic>**140| Field | Value |141(repeat per post)142 143Rules: <each rule: held | broken at slot n, why> · Persona gaps: <weeks, missing roles, or none> · Duplicates: <topic → match, or none>144 145<tab-separated block in a code fence: header row, then one row per post>146Learnings applied: <list, or none>147```148 149Then the handoff block.150 151## Hand off152 153```handoff154skill: content-calendar155version: 1.0.0156verdict: READY157coverage: complete158for: <source name>159outputs:160 posts: 9161 weeks: 3162 start: <YYYY-MM-DD>163 rows: [<week, date, topic, pulled_from, hook>]164 anchors: 1165next: [founder-post]166open: []167```168 169`founder-post` takes one row at a time: it re-reads the source from `pulled_from` and keeps the row's hook word for word. At `AWAITING PICK`, `next` is `[]`.170 171## Failure modes172 173| Problem | Do this |174| --- | --- |175| The source is a sales page or a list of features | Few or no nuggets pass point 6 or 7; say so, return `NO TOPICS` if none |176| The reader says "pick for me" | Rank the topics (anchor first, then strongest consequence) and show the ranking at the checkpoint; still wait for a yes |177| A topic is already in the existing calendar | Flag it with the match; the reader decides whether to keep it |178| The picks are all one persona | Build the calendar, report the gap, and point to the topics at the checkpoint that would fill it |179| The reader names 3 topics as an arc | Keep them back to back in the order problem, reframe, proof, and tag each with its role |180| A figure in a hook has no source | Take the figure out of the hook; keep the topic |181 182## Influenced by183 184- Richard van der Blom's research on LinkedIn reach, which found the best posting frequency has dropped to 2 to 4 posts a week: https://creatorscience.com/podcast/richard-van-der-blom-2/185- SHAKEUP's social pipeline: an operator lens and a buyer lens on the same source, a strict quality bar for what counts as a topic, an anchor per chapter held to a narrative test, framing on every topic before the pick, and a paste-ready block instead of a push.