gooseworks-ai/goose-skills / skills/marketing/run-organic-social
run-organic-social
VerifiedRuns a brand's organic social: how each platform works now, a strategy per platform and two weeks of posts in the brand's or a person's real voice. Use when a marketer says "run organic social for my brand", "fill my social calendar", "plan two weeks of posts", "grow my X / LinkedIn / Instagram / TikTok", "ghostwrite my founder's posts", "which platforms should we be on?", "plan the social for our launch", "a TikTok or Reel script to shoot on my phone", "tear down @x's account", or "my accounts went quiet". The marketer posts by hand; nothing is posted or scheduled. Not for a standalone post or repurposing one piece — use make-content.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 23.1 KB |
| references/calendar-craft.md | markdown | 11.5 KB |
| references/graphics.md | markdown | 5.8 KB |
| references/platform-native.md | markdown | 3.6 KB |
| references/platform-strategy.md | markdown | 8.7 KB |
| references/post-checks.md | markdown | 10.6 KB |
| references/report-shape.md | markdown | 6.6 KB |
| references/run-record.md | markdown | 8.5 KB |
| references/scheduled-refill.md | markdown | 6.9 KB |
| references/slot-anatomy.md | markdown | 7.6 KB |
| references/voice-card.md | markdown | 4.4 KB |
| skill.meta.json | config | 7.9 KB |
| tests/test-by-hand.md | markdown | 9.8 KB |
references/post-checks.md
View on GitHub# Post checks: share-worthiness, AI-tells, claims, disclosure, trends
Load at the check step. Every draft passes all five sections, and a failing draft is remade before anything is shown.
## 1. "Would a human share this?" (fail on any)
1. **Specific:** at least one detail only this brand or person could say: their product in use, a signed-off number, a buyer's own phrase from the voice-of-customer bank (press quotes are not buyer words), a real place, a real mistake.
2. **Swap test:** it no longer works with a competitor's name in.
3. **Share motive named:** makes the sharer look smart · useful to a specific friend · funny · "this is so me" · worth saving for later. Write the motive in the slot. No motive, no post.
4. **Hook:** the first line or first second stands alone within the platform's truncation. No throat-clearing.
5. **Payoff:** the body delivers what the hook promised.
6. **Takes a side** (point-of-view posts): says something a reasonable person could disagree with.
7. **One idea** per post.
8. **Native:** it would not look odd from a real person on that platform (references/platform-native.md). Length is within the norms.
9. **Not an ad:** no feature lists, superlatives or "we're excited". The product share stays within the mix cap.
10. **Read-aloud:** it sounds spoken, not written.
11. **Variety (calendar level):** no two posts in the window share an opener structure or a CTA, nothing repeats last run's posts (read the previous calendar), and nothing re-runs a post already live on the account (`brand_read organic.own_posts`) unless it's a deliberate re-cut.
## 1b. "Would a person actually post this?" (the founder test, fail on any)
Read each draft as the person whose name is on the account, and as someone scrolling past it.
1. **Only they could have written it.** It contains something from their own material: what they shipped, a bug, a number they can share, a screenshot, a customer moment, a view they'd argue. Generic advice ("3 things to write down before…") with the product bolted on fails.
2. **It isn't an ad.** Most posts are not about the product at all. When the product appears, it's the proof inside the story ("here's the screen where it caught this"), not the closing pitch. No "We built X so that…" endings.
3. **It sounds like them on that platform.** On X: short, specific, lowercase if they write that way, one idea, no LinkedIn cadence. On LinkedIn: a real story with a turn, not a press release. On TikTok and Reels: spoken words, not written ones.
4. **It follows a pattern that works there.** It names the example it's modelled on (slot "Modelled on"). A post with no proven pattern behind it is a guess: drop it or make it a labelled test.
5. **No engagement bait or filler questions.** No "I'll read every reply", "Agree?", "What do you think?" endings, and no weekly recap posts with empty brackets. A question post must be one the person genuinely wants answered, and specific.
6. **Every format has a reason.** A video needs something worth watching (a finished output appearing, a before/after); a text post needs a point that lands without motion. A platform "pushing" a format is not a reason on its own.
7. **Postable as written.** At most one short fill-in that only the person knows. A post that reads as a template fails.
8. **Could this make a buyer trust us less?** A bug list, an outage, a cost surprise or anything that reads like one customer's data reaching another needs a lesson and a fix the buyer cares about, or it doesn't go out. Build-in-public works when there's a lesson; a list of what broke is not one.
9. **Useful to a stranger?** (calendar level) At least 3 posts in 10 help someone who has never heard of the brand, and at most half are mainly about the product. Changelog, install and plumbing posts inform existing users; they rarely earn a follow.
A draft that fails is rewritten from the person's material, or from material found in the free pass, or it is cut. Two weak posts are worse than one good one.
## 2. AI-tell blacklist (calibrated against the voice card)
If the account's real posts use an item, it is allowed. Otherwise remove it.
- **Words:** delve, elevate, unlock, unleash, harness, empower, leverage, seamless, robust, game-changer, cutting-edge, revolutionize, transformative, supercharge, landscape, realm, tapestry, testament, embark, journey (unless literal), navigate (figurative), deep dive, "fueled by", "at the end of the day".
- **Phrases:** "In today's fast-paced world", "In the ever-evolving…", "Here's the thing", "Let's dive in", "The best part?", "Here's the kicker", "Let that sink in", "The result?", "The secret?", "Spoiler:", "Nobody talks about…", "Most people don't realize", "X isn't dead", "Say goodbye to X and hello to Y", "Whether you're X or Y", "Imagine a world where", "Ready to…?", "Remember:", "In short,", "I'm thrilled/humbled to announce", "Thoughts?", "Drop a comment below 👇", "Agree?".
- **Structures:**
- "It's not X. It's Y." reframes; "Not X. Not Y. Just Z."
- Rhetorical triads.
- Whole posts of one-line staccato paragraphs.
- Self-answered questions.
- Colon reveals.
- Symmetric listicles with bolded labels.
- Unicode bold or italics.
- Title Case Hooks.
- Evenly balanced sentence lengths.
- A summarizing moral at the end.
- Invented dialogue in quote marks.
- **Punctuation and emoji:** em dashes no more often than the card's rate (default: none). No emoji bullets (✅🚀💡👉🔥✨) unless they are in the card. No hashtag tails. Exclamation marks at the card's rate.
## 3. Claims and banned words
Claims follow the `run-organic-social-post-check` skill (`../../run-organic-social-post-check/SKILL.md`) step 4: the shared claims gate plus the social spans (world facts, product match, past posts are not proof). Remove every banned word from the voice guide and every `dont` learning. In regulated categories (beauty, health, supplements, finance, kids, alcohol) there is no treat or cure language beyond signed-off wording. Offers and prices appear only as dated in the brain; never invent a discount, drop date or giveaway.
## 4. Disclosure and rights
- **Disclosure** (paid, gifted, affiliate, employee, founder and ambassador material connections; UGC reposts and the rights ask; contests): follow the shared organic disclosure rules in `../../ad-policy-check/disclosure.md` (in the `ad-policy-check` skill). Label at the start, never buried. Never assume a repost right: draft the rights-ask message for the marketer to send.
- **Third-party IP and real people (every slot, not only trends):** never feature a character, franchise, game, show, logo, or a real person's name or likeness that the brand doesn't own. That includes fan-made or user-made items in the brand's own catalog. Only with a licence the human confirms; otherwise showcase the brand's own original items, and ask when unsure.
- **Sensitive topics** (mental health, suicide, self-harm, medical): a help or crisis resource line from a source opened this run or the brand's own standard line (else `[RESOURCE LINE NEEDED]`); no product tie-in.
- **Age-sensitive content:** don't promote mature or romance-explicit items where the account's audience may include minors.
- **Music:** TikTok brand accounts use the Commercial Music Library only. Never script a trending song for a brand account.
## 5. Trend posts
**Is it really a trend?** It must show acceleration (growing this week against last), repetition across unrelated accounts, or spread across platforms. Big lifetime views alone don't count.
**Pick one of five responses:**
1. join in;
2. borrow the format for our own topic;
3. answer the question the trend raises;
4. make an evergreen version that outlives it;
5. ignore it.
Say which, and why. Also name the horizon: react this week, plan for next month, or let it shape the quarter.
Trends come only from fresh evidence: this run's platform research (the `run-organic-social-platform-research` skill, `../../run-organic-social-platform-research/SKILL.md`), research-market's latest report (the "What people are saying" section of the newest `research/market/<D>-read.md`: trends with post-by dates; check its window line), and the inspiration library (`gooseworks:social_inspiration_search`). Every trend carries its first-seen date and a link.
**Fit test (all four must pass):**
1. The audience already knows the trend.
2. The brand adds a real twist from its niche, not just its logo on the meme.
3. The voice card allows it.
4. It can be posted before it expires, in a slot when the human is free.
**Tone risk (name it on every trend draft):**
- **Green:** format, meme or sound templates; harmless pop culture; seasonal moments.
- **Amber:** celebrities, competitor jabs, jokes about customers, industry drama, anything politics-adjacent. Give a recommendation and the specific way it could go wrong.
- **Red, never draft:** tragedy, disaster, war, elections, religion, identity, health scares, layoffs, lawsuits, anything where someone got hurt.
- **IP:** copyrighted audio, clips or characters on brand accounts.
**News as the hook:** news more than 7 days old on the posting day is never the hook (it can be context inside the post). **Seasonal posts go out when the buyer plans, not on the day.** A post helping agencies with Black Friday creative goes out before they build it (weeks ahead), not on 6 November; a shopper-facing offer posts in the days before. Check the buyer's lead time and say it in the slot.
**Expiry** = the report's post-by date when it gave one; otherwise the earlier of the platform half-life (below, counted from first seen) and the human's next free posting window. **Dated cultural moments** (Halloween, Diwali, Black Friday) are fixed slots, not trends: they carry the date as their expiry and a tone-risk tier, but no TREND flag. If it is past peak when drafted, drop it and say so. A stale trend post is worse than a skipped slot.
| Platform | Half-life |
|---|---|
| X meme or news | 24–48h |
| Threads | 1–3 days |
| TikTok sound or format | 3–7 days (if brands are already doing it, you are late) |
| IG Reels audio | 1–2 weeks |
| LinkedIn news commentary | 2–5 days |
| Dated cultural moment | The date itself |
| Pinterest | No trends; plan 45–90 days ahead |
## Check result (attached to every slot)
```
Checks — <slot id>
Share motive: … · swap test ✓ · hook within truncation ✓ · native ✓
AI-tells removed: …
Claims (signed-off <date> | scoped sign-off this run | product facts only): kept · softened · cut (would need …)
Banned words: none | removed …
Disclosure: n/a | added "…"
Trend: n/a | tone risk green/amber (…) · expires <date time tz> · recommendation: post / skip
Person: n/a | PERSON YES NEEDED (<name>, approver <who>)
```
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |