agentleFS
inklateinklate/social-skills/skills/social-crosspost/SKILL.md

social-crosspost

Skill8 starsChanged 4 months agoMITView on GitHub

Turn one story, blog post, announcement, or idea into native-feeling drafts for multiple platforms at once — LinkedIn, X (Twitter), Instagram, Facebook, Threads, Bluesky, and Mastodon — adapting tone, length, structure, hashtags, and format per platform instead of copy-pasting one caption everywhere. Use when someone says "cross-post", "post this everywhere", "post this on LinkedIn, X, and Instagram", or "adapt this for LinkedIn and X". Each draft comes with a one-line note explaining what changed for that platform and why. For turning one long source into a week of different posts, use social-repurpose instead. Reads social-context.md for brand voice, audience, and default platform setup so every draft sounds like the same person speaking each platform's dialect.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 5 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read social-context.md (also check .agents/social-context.md) before drafting.

    line 31
  2. Very specific:

    If the file is missing, offer to run the social-context skill first — but do not block.

    line 38
  3. Very specific:

    Confirm target platforms: Use the user's list if given.

    line 44
  4. Distinctive:

    Voice-check the set: Read all drafts in a row and hold them to three tests:

    line 47
  5. Distinctive:

    No verbatim repeats — kill any phrase that appears word-for-word in three or more drafts.

    line 49
  6. Distinctive:

    Every draft must contain the core claim.

    line 58
  7. Distinctive:

    Never pad a short-form draft to feel "complete" — Bluesky at 180 chars beats Bluesky at 300.

    line 60
  8. Distinctive:

    No draft may trip the Never list.

    line 65

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Given one source — a URL, pasted text, or a raw idea — produce a native draft per platform, each shaped for how that platform actually reads, with a one-line rationale per draft.

Context

Read social-context.md (also check .agents/social-context.md) before drafting. You need:

  • Brand voice: register, person ("I" vs "we"), emoji policy, banned phrases
  • Audience per platform, if specified — the LinkedIn reader and the Bluesky reader are rarely the same person
  • ## Platforms: the default platform list when the user does not name targets
  • The Never list: red lines to keep out of every draft, since one message goes to many public feeds

If the file is missing, offer to run the social-context skill first — but do not block. Ask two or three quick questions inline (Which platforms? Personal voice or company voice? Anything off-limits?) and proceed.

Workflow

  1. Ingest the source. Fetch the URL or read the pasted text in full — the best social material is often a buried aside, not the headline. If it is only an idea, ask one clarifying question at most, then work with what you have.
  2. Extract the ONE core claim plus 2–3 supporting points. Write these down and show them before drafting. Every platform draft argues the same claim; what varies is how. If you cannot state the claim in one sentence, the source is not ready — say so and propose the sharpest available angle. Also harvest the receipts: any number, quote, or before/after in the source, because those survive adaptation better than prose.
  3. Confirm target platforms. Use the user's list if given; otherwise default to ## Platforms from social-context; otherwise ask. Do not silently produce all seven — an unwanted Mastodon draft is noise, not thoroughness.
  4. Draft each platform natively, from the claim — not from another draft. These are siblings, not truncations of each other. Load references/platform-rules.md and draft each target to its length, structure, hashtag, and link rules there. Keep the receipts (numbers, quotes, before/after) — they survive adaptation better than prose — and give every platform its own hook, since reusing one first line wastes the variable that matters most.
  5. Write the per-draft rationale. One line under each draft: what you changed for this platform and why ("compressed to the stat because X rewards one sharp number", "led with the customer story because LinkedIn's fold buries anything slower"). The rationale teaches the user the platform, not just this post.
  6. Voice-check the set. Read all drafts in a row and hold them to three tests:
    • Same person, different rooms — if any draft could not have been written by the voice in social-context, revise it.
    • No verbatim repeats — kill any phrase that appears word-for-word in three or more drafts; siblings, not clones.
    • Claim intact — every draft still argues the core claim from step 2; a draft that drifted into a different point gets rewritten, not kept.
  7. Check every draft against its platform rules. Using references/platform-rules.md, verify each draft on every axis — length, structure, hashtags, links — and fix violations: count characters on X, Threads, Bluesky, and Mastodon and trim to fit by cutting the weakest middle clause, never by amputating the ending. Screen every draft against the Never list too.
  8. Flag gaps honestly. If a platform is a poor fit for this story (a B2B pricing memo on Instagram), include the draft but say so in its rationale rather than forcing false enthusiasm — the user may still want it, but they should not be surprised when it underperforms.

Quality bar

Per-platform limits and drafting notes live in references/platform-rules.md — check every draft against its platform's row there, not against memory. Across the whole set:

  • Every draft must contain the core claim; no draft may contradict another.
  • Numbers and concrete nouns survive adaptation; adjectives do not have to.
  • Never pad a short-form draft to feel "complete" — Bluesky at 180 chars beats Bluesky at 300.
  • Hooks differ per platform: reusing the same first line on LinkedIn and X wastes the one variable that matters most.
  • Hashtags are platform culture, not decoration: Instagram expects them, X and Threads punish them, Mastodon wants them CamelCase.
  • Emoji follow the social-context policy; when in doubt, LinkedIn and Mastodon get fewer, Instagram and Threads tolerate more.
  • Thread decisions are structural: only thread on X when there are 3+ distinct beats — a two-tweet thread is a long tweet that lost its nerve.
  • No draft may trip the Never list.

Deliverable

Return, in order:

  1. A two-line summary: the core claim in one sentence, and the platforms covered.
  2. One section per platform: a heading, the ready-to-copy draft in a fenced block, and the one-line rationale beneath it.
  3. Character counts on every draft with a hard cap (X, Threads, Bluesky, Mastodon), shown as (214/280).

Nothing else — end there.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.