원클릭으로 Manus에서 모든 스킬 실행

social-copy

스타109

포크10

업데이트2026년 6월 14일 22:03

Fast, bulk social post COPY — a ready-to-paste description + a trending-hashtag set for a finished reel / image / carousel, shaped per platform (TikTok hook + 3-5 tags · Reels caption + 10-15 tags · Shorts ≤40-char title) in ONE pass. The last-mile of every published piece: source-grounded from the project's units / brief / the reference reel's on-screen text + caption (not a generic guess), in the niche's native voice (meme / brainrot register for the aura format, professional for a brand ad), in the TARGET-AUDIENCE language (chat language ≠ post language — offer EN + the audience language). Bulk mode runs across N units / a whole batch and emits one copy block per item. Persists into unit.json via `ralphy unit caption` (append-only). NOT video-subtitle SRT — that is `ralphy generate captions`. USE WHEN the user asks to "write a caption", "description + hashtags", "make a description", "captions for the batch", "post copy", "social copy", "what should I write under this reel", "give me hashtags", "caption th

설치

Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.

Manus에서 실행

출처

alecs5am

alecs5am/ralphy

GitHub 저장소 열기 Creator 저장소 보기

다운로드

Manus에서 실행

관련 직업SOC

SOC 직업 분류 기준

기술 작가예술, 디자인, 엔터테인먼트, 스포츠, 미디어직·SOC 27-3042

SKILL.md

readonly

이 저장소의 다른 Skills

같은 저장소

social-publish

alecs5am/ralphy

Publish or schedule a finished Ralphy Unit through the active workspace's saved Postiz connection. Prepares the target-shaped title, description, body, CTA, hashtags, thread parts, and required Postiz settings, then calls `ralphy publish` and records provenance for later analytics. USE WHEN the user asks to "publish this", "post it now", "schedule this post", "send this to Telegram", "post this thread to X", "publish the reel", or "put this on YouTube/Instagram". DO NOT FIRE for copy-only requests with no posting intent; use social-copy instead.

2026-07-13109

dev-loop

alecs5am/ralphy

Run an autonomous, issue-driven development loop over Ralphy itself — the same cadence as a focused maintainer session: pick the open `notes/issues/`, order them by dependency, then for each issue dispatch a scoped sub-agent, review its diff, run the real gates, and commit + push to main before moving to the next. Sequential by default (one agent at a time) so changes never collide on shared files. Default scope is ALL open issues with a safety limit of 25; the user can widen the scope or raise the limit in their prompt. Pauses for the user only at genuine decision forks and before any paid generation or irreversible/outward-facing action. USE WHEN the user types `/dev-loop`, says "work through the issues", "run the dev loop", "grind the backlog", "implement all the open issues", "keep going through notes/issues until done", or after `/dev-issues` has filed a batch they now want executed. DO NOT FIRE for a single issue (just implement it), for content/video production, or when there are no open issues to exec

2026-07-13109

dev-publish-template

alecs5am/ralphy

Publish a finished Ralphy unit and its reusable blocks to the public library. Use when the user asks to publish a template, unit, style, recipe, or asset to the library. The CLI prepares the entity bundle; the sibling ralphy-web repository owns the committed library and upload script.

2026-07-13109

dev-release

alecs5am/ralphy

Cut a Ralphy CLI release across GitHub Releases, Homebrew, and npm. Use when the user asks to release, tag, publish, or prepare release notes for the CLI. Public documentation and the marketing site live in sibling repositories and are synchronized separately when their content is affected.

2026-07-13109

fixer

alecs5am/ralphy

The eval-to-repair loop — the consumer of the evaluator's report (#409). Reads the eval output (eval-deep-vision.json's `what_to_redo` first, else eval.json `findings[]`), runs `ralphy project repair-plan <id>` to classify every finding by owner (art-director / scenarist / editor) and order it by severity, presents the ordered plan grouped by owner, gets the user's approval, then applies targeted fixes through the EXISTING `ralphy` verbs (art-director regen / scenarist rewrite / editor recut), re-renders, re-evals, and compares the old vs new verdict. HARD GATE: no paid model regeneration runs until the user approves the plan (or previously opted into batch repair). The deterministic `repair-plan` step itself makes ZERO model calls and is safe to run unprompted. USE WHEN the user says "fix the eval", "repair this render", "make it shippable", "fix the issues / problems / findings", "address the eval", "now fix it", "make the fixes", or hands off a failed / warn eval and asks what's next. ALSO FIRE right after

2026-07-13109

templater

alecs5am/ralphy

The full extract → classify → blueprint → de-dup → publish pipeline for a finished `.ralphy/workspaces/<ws>/projects/<id>/`. Reads its `units/*/unit.json` (#069) as the Unit source of truth, then factors the project into ALL FIVE entities — Unit + the four typed blocks Template (structure), Style (look), Recipe (effect/treatment), Asset (reusable media). Applies the recipe-vs-tag split (#082/#083): a candidate is a Recipe only if it carries an extractable artifact (ffmpeg filtergraph / HyperFrames snippet / encode recipe / prompt technique); otherwise it is a Tag on the Unit. Captures a per-unit Blueprint (#080), de-dups every block against the live library FIRST, then prints the exact ordered `publish-entity.ts` commands (dry-run → --push). Publishing itself is the #056 path. USE WHEN the user says any of: "save this as a template", "turn the project into a template", "templatify <project-id>", "extract a template from <project>", "decompose this project into blocks", "what units/blocks did this project prod

2026-07-13109

name	social-copy
namespace	user
description	Fast, bulk social post COPY — a ready-to-paste description + a trending-hashtag set for a finished reel / image / carousel, shaped per platform (TikTok hook + 3-5 tags · Reels caption + 10-15 tags · Shorts ≤40-char title) in ONE pass. The last-mile of every published piece: source-grounded from the project's units / brief / the reference reel's on-screen text + caption (not a generic guess), in the niche's native voice (meme / brainrot register for the aura format, professional for a brand ad), in the TARGET-AUDIENCE language (chat language ≠ post language — offer EN + the audience language). Bulk mode runs across N units / a whole batch and emits one copy block per item. Persists into unit.json via `ralphy unit caption` (append-only). NOT video-subtitle SRT — that is `ralphy generate captions`. USE WHEN the user asks to "write a caption", "description + hashtags", "make a description", "captions for the batch", "post copy", "social copy", "what should I write under this reel", "give me hashtags", "caption this drop". TRIGGER (EN): "write a caption", "caption for this reel", "description and hashtags", "social copy", "post description", "captions for the batch", "trending hashtags for <niche>". See body for ALSO FIRE / DO NOT FIRE / HARD INVARIANTS.

social-copy

The fast, batchable, on-voice social post copy layer — description + trending hashtags for a finished piece, shaped for TikTok / Reels / Shorts in one invocation. The skill DRAFTS; ralphy unit caption PERSISTS the copy into unit.json.

ALSO FIRE

The user just rendered / formed a unit and asks "and now? what do I write under it?" (any language).
The user drops a niche + "give me trending hashtags" with no caption ask — still this skill (hashtag bank + format shaping).
A batch of reels just finished and the user wants copy for all of them → bulk mode.

DO NOT FIRE

The user wants aligned-to-VO subtitles / SRT burned into the video — that is ralphy generate captions (a different surface). This skill is the social POST copy.
The user wants the script / scenario itself — that is the scenarist playbook.
The user wants to publish to the library — that is templater / dev-publish-template; caption is an input that path can carry.

Hard invariants

Target-audience language drives the copy, NOT the chat language. Same rule as intake: if the audience is German, the post is German even if the user is chatting in English / Russian. Offer EN + the audience language. Pass --language <lang> to the verb.
Source-ground every copy block. Pull from the unit's title / blurb / tags / provenance, the project brief, and — when present — the reference reel's meta.info.json / transcript / on-screen text. Copy must describe the ACTUAL content, never a generic guess.
Match the niche's native voice. Meme / brainrot register for the aura format; professional for a brand ad. The niche is read from the matched #412 content-mode / the unit's tags / provenance, then carried into the draft.
Hashtags come from the maintained bank, not invented. The per-niche trending-tag bank lives at cli/lib/social/hashtag-bank.ts (niche spine + format tags + broad-reach, deduped + ordered). Trend tags rot — see the staleness note below.
Append-only on the persisted caption. ralphy unit caption never clobbers an existing caption: re-drafting requires --force, which archives the prior into caption_versions. AGENTS.md invariant #14.

Platform shaping (one invocation, all three)

Platform	Shape	Tags
TikTok	one punchy hook line (1 sentence, scroll-stopping)	3-5 inline tags
Reels	a fuller caption (2-4 sentences, on-voice, soft CTA)	10-15 tags
Shorts	a title, ≤ 40 chars	tags optional

The drafter writes the three copy bodies; the hashtag bank supplies the merged tag set (the verb caps it ~18, niche-first → format → broad-reach).

The bulk-mode loop

For N units / a batch, do NOT hand-write each. Run once across all units:

ralphy unit caption <project> --bulk --language English

This drafts + writes one caption block per unit (skipping any that already have one unless --force). Single unit:

ralphy unit caption <project> <slug> --language English --niche aura

Useful flags: --niche <hint> (overrides the auto-detected register / tag spine), --brief "<on-screen text / source caption>" (extra grounding), --force (re-draft, archive prior).

Source grounding

Before drafting, gather:

The unit's unit.json — title, blurb, tags, provenance (style / template / recipes / assets).
The project brief / STORYBOARD.md / PRODUCTION_PLAN.md if present.
The reference reel's own caption + on-screen text (we store the source reel + its meta.info.json / transcript). The copy should echo the actual hook, not a generic line.

The verb folds the unit fields + --brief into the draft automatically; pass --brief when the strongest grounding (the source reel's caption / on-screen text) lives outside unit.json.

The trending-hashtag bank + staleness

cli/lib/social/hashtag-bank.ts maps niche → curated tag spine (aura, analog-horror, ps1-horror, tech-explainer, ugc-ad, unboxing, beauty, fitness, food, saas, general) plus format tags (reel / short / tiktok / carousel / …) and a stable broad-reach set (#fyp etc.).

Trend tags rot. The bank stamps LAST_REVIEWED; the verb's JSON output reports bank_stale: true once it is older than 60 days. When stale (or for a hot niche), refresh BEFORE shipping:

light path — run the researcher skill / a trend scrape on the niche + platform, read the current top tags, update the bucket in hashtag-bank.ts, bump LAST_REVIEWED.
manual path — open the niche's top recent posts and copy the tags they rank under.

Broad-reach tags rarely rot; format + niche buckets drift on a ~monthly cadence.

Worked example (one reel → TikTok / Reels / Shorts block)

A finished aura-moment-001 unit (PS1-horror "aura" meme reel, format video, tags ["aura","ps1core","meme"]), audience English:

ralphy unit caption aura-proj aura-moment-001 --language English

Drafts (niche auto-resolves to aura) and writes into unit.json:

{
  "caption": {
    "platform": {
      "tiktok": "POV: your aura hits 999,999 and the room goes PS1",
      "reels": "When the aura is too strong the graphics downgrade to PS1. Caught it on camera before reality re-rendered. Watch till the end. Follow for more aura moments.",
      "shorts": "Aura: 999,999 (PS1 mode)"
    },
    "hashtags": ["#auramoment","#aura","#aurapoints","#ps1core","#brainrot","#meme","#sigma","#edit","#reels","#reelsinstagram","#instareels","#fyp","#foryou","#foryoupage","#viral","#trending"],
    "language": "English",
    "niche": "aura"
  }
}

To also offer the audience-language variant, re-run with --language <lang> --force (the EN block is archived into caption_versions).

CLI cookbook

Draft + persist one: ralphy unit caption <project> <slug> [--language <lang>] [--niche <hint>] [--brief "<text>"]
Bulk across the project: ralphy unit caption <project> --bulk [--language <lang>]
Re-draft (append-only): add --force (prior caption → caption_versions).
Read it back: ralphy unit show <project> <slug> (the caption field).

Never write copy straight into unit.json by hand — the verb owns the write (logs the LLM draft to generations.jsonl, merges the bank, enforces append-only).