Skip to content

ericrisco/rsc-harness / skills/shortform-packaging

shortform-packaging

Verified

Use when a vertical short is shot or scripted and you need the upload-form copy and cover that win the feed — the hook line, the first-frame on-screen text, the search-led caption, a tight hashtag set, and the cover frame; it learns from what performed via the 02-DOCS log. NOT inventing the idea (that is `shortform-ideation`), NOT scripting or directing the cuts (that is `video-shorts`), NOT executing the edit (that is `shortform-editing`), NOT scheduling the post (that is `social-publisher`).

FileKindSize
SKILL.mdskill md14.1 KB
evals/README.mdmarkdown653 B
evals/cases.yamlconfig3.1 KB
references/package-templates.mdmarkdown5.4 KB
scripts/verify.shexecutablescript7.3 KB

references/package-templates.md

View on GitHub
# Package templates & pattern library

Depth offloaded from `SKILL.md`. Use this when you need the full package file format,
the hook/on-screen-text pattern library, the cover-frame safe-area in pixels, or the
filled `02-DOCS` feedback-log schema.

## The package file format

Emit one markdown file per short under `02-DOCS/wiki/shortform/`. The `shortform/`
wiki tree is an **OKF v0.1 bundle** shared with the `tiktok-api` and
`shortform-strategy` siblings, so the file leads with YAML frontmatter (a non-empty
`type` is the only OKF hard requirement; `title`/`tags`/`timestamp` are the recommended
surface). The labeled body below is unchanged — `scripts/verify.sh` lints exactly these
labels (English or ES/CA, one per line, content following) and ignores the frontmatter:

```text
---
type: shortform-package
title: faceless productivity TikTok — package
tags: [shortform, packaging, tiktok]
timestamp: 2026-06-02T12:00:00Z
platform: tiktok
---
Hook:
  - I deleted my to-do app. Here's why.
  - The app that runs my whole week (it's free)

On-screen text:
  - 3 apps that replaced Notion
  - I deleted my to-do app

Caption:
  3 Notion templates that replaced my to-do app — the full setup, free links below.
  Send this to the friend drowning in tabs.

Hashtags:
  #notiontemplates #productivitytips #notionsetup

Cover:
  frame: 00:07 (peak — laptop closing, face mid-grin)
  overlay: I deleted my to-do app

Feedback:
  intro_retention: 0.74
  sends_per_reach: 0.018
  saves: 312
```

### TikTok (search-led) vs Reels (send-led) — same clip, tuned

```text
# TikTok — lean into in-app search
Caption: How to batch a week of shorts in one afternoon — exact workflow inside.
Hashtags: #contentbatching #shortformvideo #creatorworkflow

# Reels — lean into the DM send
Caption: This is how I never run out of shorts. Send it to your co-founder.
Hashtags: #reelstips #contentcreator #solopreneur
```

Caption keywords and hashtags differ per platform's search audience; the hook,
on-screen text, and cover frame stay identical (shared 9:16 / 1080×1920 spec).

## Hook + on-screen-text pattern library

Each is 4-7 words, promise/tension first, and the on-screen text is the silent twin.

| Pattern | Bad | Good |
|---|---|---|
| Open loop | "Let me explain something" | "Nobody does step 3 — and it's the one that matters" |
| Number | "Some tips for you" | "5 shorts in one afternoon" |
| Contrarian | "Posting daily is good" | "Stop posting daily. Do this instead." |
| Stakes | "A mistake to avoid" | "This caption mistake kills your reach" |
| POV | "Here is my routine" | "POV: you finally batched a week of content" |
| Before/after | "It got better over time" | "0 to 10k saves with one caption change" |

Rule: whichever you pick, the **on-screen first-frame text restates it in the same words**
for the sound-off ~60%, inside the safe zones (clear of top status bar and bottom UI).

## Cover-frame selection & safe area (in pixels)

The cover is the grid/shelf billboard, not the in-feed frame. The same 1080×1920 source
frame is cropped three different ways depending on where it surfaces:

- **Reels tab / Shorts shelf:** shows the full **9:16 frame, 1080×1920** — no crop.
- **Profile grid:** crops to **3:4, 1080×1440**. Instagram moved the grid from 1:1 to 3:4
  in 2025, lopping the **top ~240px and bottom ~240px** off the 1920-tall frame
  (the surviving band is roughly **y=240 to y=1680**). Sources: postfa.st "Instagram Reels
  Size", Buffer 2026.
- **Main feed:** crops to **4:5, 1080×1350** (a different, taller centre band again).

You cannot author one band that pleases all three crops separately — so author for the
**intersection**: the **1:1 centre square, 1080×1080** (vertical centre, roughly
**y=420 to y=1500**) sits inside every one of the three crops above. Keep all overlay
text in that square and it survives the grid, the feed, and the tab. Then also clear the
platform UI overlays (bottom ~250px: caption, action buttons, audio tag; top ~120px:
status bar). Practical safe band for overlay text: centred, between y≈600 and y≈1350.

Checklist:

- [ ] Frame shows motion / high contrast / a face mid-expression / the payoff
- [ ] Overlay is 3-6 words, same words as the hook line
- [ ] Overlay sits inside the 1:1 centre square (the intersection that survives the 3:4 grid, 4:5 feed, and 9:16 tab crops)
- [ ] Text clears top ~120px and bottom ~250px UI bands

## 02-DOCS feedback-log schema

Append one entry per shipped short under `02-DOCS/raw/shortform/`. Key on distribution
metrics, never likes. Read the log before writing the next package and mirror the
winning shape.

```yaml
- shipped: "2026-06-02 / faceless productivity TikTok"
  platform: tiktok
  hook: "I deleted my to-do app. Here's why."
  on_screen_text: "3 apps that replaced Notion"
  caption_lede: "3 Notion templates that replaced my to-do app"
  cover_overlay: "I deleted my to-do app"
  hashtags: ["#notiontemplates", "#productivitytips", "#notionsetup"]
  intro_retention: 0.74    # past-3s rate; bar is 0.70
  sends_per_reach: 0.018   # DM shares — the share-economy currency
  saves: 312               # weighted above likes (stronger intent signal)
  variant_tested: "number-led vs question on-screen text"
  won: "number-led lifted intro_retention 0.74 vs 0.61; carry forward"
```

Fields the next package reads first: `won`, `intro_retention`, `sends_per_reach`. If the
last winner was a number-led hook, start the next set there and test against it.
rea-changelog-updatemorluto/reaPrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
academy-guideanthropics/skillsStop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
claude-apianthropics/skillsReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).Stars: 198kUpdated Claude CodeCodexUnverifiedInstall
doc-coauthoringanthropics/skillsGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.Stars: 198kUpdated Claude CodeCodexVerifiedInstall