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-adgit 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-ad)<a href="https://agentmods.dev/skills/inklate/social-skills/social-ad"><img src="https://agentmods.dev/badge/skills/inklate/social-skills/social-ad/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-ad"><img src="https://agentmods.dev/badge/skills/inklate/social-skills/social-ad.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 | $0.00163 | $0.01837 |
| Opus 5 | $0.00081 | $0.00919 |
| Sonnet 5 | $0.00033 | $0.00367 |
| Haiku 4.5 | $0.00016 | $0.00184 |
Grade A, and why
social-ad 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 12d 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 — 93 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Produce placement-ready ad copy variant sets — an angle matrix crossed with every copy part each platform needs, plus a one-line creative brief per variant.
Context
Read social-context.md at the project root (also check .agents/social-context.md) for brand voice, audience, product positioning, and banned phrases (proof points come from the offer and landing page in step 1). If it's missing, offer to run the social-context skill first, but don't block — ask 2–3 quick inline questions and proceed:
- What's the offer, the price, and the single strongest proof point?
- Who's the audience, in one sentence?
- Any voice rules — words you never use, no exclamation marks, no "unlock"?
Workflow
- Ingest the offer. Get the landing page URL or offer description and read it closely. Extract four things: the concrete promise, the price/commitment, the strongest proof point (a number, a named customer, a guarantee), and where the CTA actually sends people. If the landing page's promise and the user's framing disagree, flag it before drafting — ad-to-page mismatch kills conversion and quality scores, and no copy fixes it.
- Ask audience temperature. Cold, warm (engaged but never bought), or retargeting (visited or abandoned)? This is not optional — the copy is structurally different:
- Cold: name the problem before the product; the reader doesn't know they're shopping yet.
- Warm: lead with the offer and the proof; they know the category, sell the difference.
- Retargeting: reference the visit ("Still thinking it over?") and lean on objection-flips and risk-reversal (trial, guarantee, cancel-anytime). If the user wants multiple temperatures, treat each as its own variant set — don't average them into mush.
- Confirm placements. Which of Meta (Facebook/Instagram feed), LinkedIn (sponsored content), X, TikTok? Only draft what's requested; each placement is real work, not a find-and-replace.
- Build the angle matrix. For this specific offer, write one crisp sentence per angle before drafting anything:
- Pain — the ongoing cost of the status quo, in the audience's own units (hours, dollars, missed deals).
- Aspiration — the after-state, concrete enough to picture, not "transform your workflow".
- Social proof — a real number or a real name from context or the landing page. Never invented; if none exists, say so and drop the angle.
- Objection-flip — name the #1 objection out loud and answer it head-on ("Yes, another tool. This one deletes three.").
- Honest urgency — a real deadline or capacity limit only. If none exists, replace this angle with a second pain or proof variant. Never fake it.
- Draft per placement. For each placement, write 3–5 variants, each from a different angle, each containing every copy part the platform uses (see Quality bar). Front-load ruthlessly: the visible-truncation point, not the hard limit, is your real budget for the first idea — a hook that dies at "…more" was never a hook.
- Write the creative brief line. One line per variant telling the designer exactly what the visual is: subject, composition, text overlay if any. Concrete ("split screen: messy spreadsheet vs. clean dashboard, overlay '4 hours → 20 minutes'"), never mood-board vapor ("something clean and modern"). The brief should visualize the angle, not just the product.
- Compliance pass. Check every variant against the rules that get ads rejected or accounts flagged:
- No second-person call-outs of sensitive attributes — health conditions, financial hardship, religion, ethnicity, age. "Struggling with debt?" fails Meta review; "Debt doesn't have to be permanent" passes. Rewrite from "you are X" to "X exists / X is solvable".
- No fake scarcity, invented countdowns, or "only 3 left" that isn't true.
- No unverifiable superlatives ("the #1 tool") without a citable source.
- Every claim in the ad must appear on, or be supported by, the landing page.
- Pick the CTA button per variant. The button is part of the copy, chosen from each platform's actual vocabulary (see Quality bar). Default mapping: "Learn More" for cold, "Sign Up" / "Get Offer" for warm and retargeting, "Shop Now" only for direct purchase, "Download" / "Request Demo" when that's literally the next step.
- Label everything. Tag each variant
[placement / angle / temperature]so when performance data comes back, the user can map winners and losers to the matrix and commission the next round intelligently. - Verify against the Quality bar. Before handing off, count every copy part of every variant against its placement's real limits in the Quality bar — count, don't eyeball — and check every variant against the Additional bars. Fix any overflow or miss; a variant that overflows its placement is not placement-ready.
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.
- 12d ago First seen · 93 lines · 163 tokens per session scan A f08097b7eab3
social-ad is a skill published in the GitHub repository inklate/social-skills (6 stars, last pushed 1mo ago), licensed MIT. It adds 163 tokens to every session and 1,837 once invoked, about $0.0008 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…
chat-pet-sprite-creation
Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.
cpu-profile-analysis
Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…
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…