Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add inklate/social-skills --skill social-carouselgit clone --depth 1 https://github.com/inklate/social-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/inklate/social-skills/social-carousel)<a href="https://agentmods.dev/skills/inklate/social-skills/social-carousel"><img src="https://agentmods.dev/badge/skills/inklate/social-skills/social-carousel/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/inklate/social-skills/social-carousel"><img src="https://agentmods.dev/badge/skills/inklate/social-skills/social-carousel.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00148 | $0.01668 |
| Opus 5 | $0.00074 | $0.00834 |
| Sonnet 5 | $0.00030 | $0.00334 |
| Haiku 4.5 | $0.00015 | $0.00167 |
Grade A, and why
social-carousel scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Given source material or a topic, produce a full carousel script — cover hook, 6–10 content slides, recap, CTA — with a design note and image prompt per slide.
Context
Read social-context.md (also check .agents/social-context.md) before drafting. You need:
- Brand voice: vocabulary, capitalization habits, whether slides may joke
- Audience: expertise level, which determines how much a single slide can assume
- The Never list, if present: red lines that disqualify a slide or a caption
social-context.md does not carry visual identity, so ask the user directly for colors, type feel, and illustration-vs-photo — one question alongside the platform question in step 1.
If the file is missing, offer to run the social-context skill first — but do not block. Ask two or three quick questions inline (Instagram or LinkedIn? Who reads this? Any visual style words — minimal, bold, hand-drawn?) and proceed.
Workflow
- Ask the platform first. Instagram (1080×1350 portrait, tap-through, big type) and LinkedIn (document/PDF, often landscape or square, denser text tolerated) are different media. Do not draft a generic script and relabel it. If the user already named the platform, skip ahead.
- Extract one narrative arc from the source. A carousel carries exactly one argument: problem → tension → resolution, or claim → evidence → payoff. If the source contains three arguments, pick the strongest and say which two you dropped.
- Write the cover slide. It is a hook, not a title: eight words or fewer of large text, optionally a small kicker line. "5 onboarding mistakes" is a title; "Your onboarding is losing users on day one" is a cover. The cover must work as a standalone feed image, because that is how most people will encounter it.
- Outline the content slides before writing them. One idea per slide, in escalating order — strongest point second-to-last, never first. Instagram: 6–8 content slides. LinkedIn: 8–10 is fine; readers expect more substance in a document post. Show the outline as one line per slide and let the user cut or reorder before you invest in full drafts.
- Draft each content slide at ≤25 words. Constraints that keep slides swipeable:
- Self-contained: each slide parses without its neighbors, because screenshots travel alone.
- Pull-through: end slides on tension or an open loop where the material allows it.
- Front-load the slide's noun; no slide begins with "Also", "And", or "Another thing".
- One number or one image concept per slide — a slide that needs both is two slides.
- Write the recap slide. Compress the arc into 3–5 checklist lines the reader can screenshot. This is the most-saved slide; make it worth saving.
- Write the CTA slide. Instagram: ask for a save or a share ("Save this for your next launch"), plus follow. LinkedIn: ask a discussion question that invites comments, plus follow. One ask primary, never three equal asks.
- Attach a design note to every slide. Three parts, all concrete:
- Layout — text position, alignment, margins ("headline top-third, left-aligned, generous bottom margin").
- Visual metaphor — the one image or diagram that carries the idea ("iceberg with 'signup' above the waterline, 'activation' below").
- Text placement — what is large, what is small, and the single emphasis word or number that gets the accent treatment. Keep every note executable by someone who never read the source: "left-aligned headline top-third, iceberg illustration lower-right, the phrase 'day one' in accent color".
- Attach an image-generation prompt to every slide that a designer or image model can run as-is:
- Subject and composition ("flat illustration of a doorway with a welcome mat, centered, seen straight-on")
- Style and palette hint, consistent across all slides — one visual system, not ten
- The explicit instruction to leave clear negative space where the slide text sits
- Swipe-check the whole script. Read only the slide texts in order, ignoring the design notes. The argument must survive on copy alone; if it does not, fix the copy, not the visuals. Then check the reverse: would the visuals alone tell roughly the same story? Finally, check the full script and the caption against every row of the Quality bar and against the Never list — fix any slide or caption that fails a row or trips a red line.
- Write the accompanying caption or post text last. Its first line is a second hook (do not repeat the cover verbatim); it adds one piece of context the slides omit, then hands off: "swipe through".
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 90 lines · 148 tokens per session scan A 13966d3a05cf
social-carousel is a skill published in the GitHub repository inklate/social-skills (6 stars, last pushed 1mo ago), licensed MIT. It adds 148 tokens to every session and 1,668 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
local-ai-agents
Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
insight-error-page
Write or audit an insight-kind error page for the Next.js dev overlay. Use when creating a new errors/ .mdx page, auditing an existing one, or checking that a page matches the framework fix cards. Covers page structure, title alignment, FixCard cards with Copy prompt button, code snippets, terminology verification…
next-cache-components-optimizer
Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…
next-partial-prefetching-adoption
Turn on Partial Prefetching in a Next.js app and work through the insights it surfaces. Use when the user wants to enable or adopt Partial Prefetching, flip the partialPrefetching flag, opt routes in with export const prefetch = 'partial', audit Link prefetch={true} behavior, preserve existing prefetched UI with…