Ad Copy Writer

Create, generate, convert, and polish ad copy, marketing copy, product copy, landing page copy, headline variants, CTA variants, and promotional messaging th...

Security Scan

The skill's requirements, scripts, and runtime instructions are consistent with an ad-copy client for the WeryAI chat API and request only the expected Node runtime and a single service API key.

Purpose & Capability

Name/description match the implementation: scripts call a WeryAI chat completions endpoint to produce ad/marketing copy. Declared requirements (node, WERYAI_API_KEY) are appropriate and proportional to the stated purpose.

Instruction Scope

SKILL.md confines runtime behavior to collecting copy parameters and invoking local node scripts that call the WeryAI API. It does not instruct reading unrelated system files. Note: the code honors optional env vars (WERYAI_BASE_URL, WERYAI_REQUEST_TIMEOUT_MS) that are not listed in requires.env — these are configuration knobs that can change the network target/timeout.

Install Mechanism

No install spec; scripts are bundled and executed with the host Node.js runtime. No external downloads or installers are requested by the package itself.

Credentials

Only a single primary credential (WERYAI_API_KEY) is required, which matches the use of a third‑party text generation API. Minor optional env vars (base URL, timeout) are present but not required; they are reasonable for configuring the client.

Persistence & Privilege

always is false and the skill does not request permanent agent-level privileges or modify other skills. It runs as an on-demand Node script and does not persist extra system configuration.

Scan Findings in Context

[system-prompt-override] unexpected: Scanner flagged a possible system-prompt override pattern in SKILL.md (e.g., directives like 'Do not explain your reasoning'). In context these are normal instructions for copy output shape rather than an attempt to subvert agent controls, but the flag is worth noting because the skill's instructions include explicit conversational behavior directives.

Assessment

This skill runs local Node scripts that send your input and the WERYAI_API_KEY to a WeryAI chat completions endpoint (default https://api.weryai.com). Before installing, confirm you trust WeryAI and that the API key you provide has the intended scope and billing account. If you want to inspect what will be sent, use the documented --dry-run mode to preview the request body without making network calls. Be aware you (or an autonomous agent using this skill) will transmit copy briefs and any sourceText to the external API. Also note optional env vars (WERYAI_BASE_URL) can redirect requests — only set that if you intend to target a non-default endpoint. The pre-scan prompt-injection flag points to strong output directives in SKILL.md (which are likely benign here), but you should avoid giving the skill secrets or unrelated credentials, and only provide the WERYAI_API_KEY if you accept that data will be sent to the external service.

Write ad copy, marketing copy, product copy, headline variants, CTA variants, and promotional messaging that are concise, persuasive, and channel-aware. Use this skill when the user wants to draft, rewrite, or polish persuasive copy for ads, landing pages, launches, or campaigns.

Prerequisites

• WERYAI_API_KEY must be set before calling the API.
• Node.js >=18 is required because the runtime uses built-in fetch.
• Real runs use the WeryAI chat completion API and may consume credits.

Example Prompts

• Write 5 ad copy variations for a productivity app launch.
• Turn this product brief into landing page copy with a stronger CTA.
• Rewrite this ad so it sounds less generic and more benefit-driven.
• Draft short paid social copy for a skincare product in English.

Quick Summary

• Main jobs: ad copy writing, marketing copy, landing page copy, headline variants, cta variants, promotional rewrite
• Default model: GPT_5_4
• Main optional controls: product, brand, audience, tone, format, cta, keywords, mustInclude, avoid
• Main trust signals: dry-run support, model lookup, channel-aware preset, direct copy-ready output

Workflow

1. Capture the product, offer, audience, channel, tone, and desired call to action.
2. If the user wants a specific ad shape such as headline set, landing page copy, CTA variants, or paid social copy, read references/domain.md and match the closest pattern.
3. Ask only for the smallest missing detail needed to write persuasive copy.
4. Use --dry-run first when you want to inspect the final payload.
5. Run node {baseDir}/scripts/write.js --json '...' and return the final copy directly.

Commands

# List available chat models
node {baseDir}/scripts/models.js

# Write ad copy
node {baseDir}/scripts/write.js --json '{
  "prompt":"Write 5 ad copy variations for a productivity app launch",
  "product":"AI productivity app",
  "audience":"busy professionals",
  "format":"paid social ad",
  "cta":"Start free trial"
}'

# Rewrite marketing copy without calling the API
node {baseDir}/scripts/write.js --json '{
  "prompt":"Rewrite this ad so it feels sharper and more benefit-driven",
  "sourceText":"...",
  "product":"skincare serum",
  "tone":"premium but clear"
}' --dry-run

Definition of Done

• The final output reads like persuasive copy rather than a memo or explanation.
• The wording matches the requested channel, audience, and CTA closely enough to use as a draft.
• If the user asked for variations, the output clearly separates them.

When Not to Use

• Do not use this for long-form articles or blog posts.
• Do not use this for pure translation without copy adaptation; use copy-translator instead.
• Do not use this for general brainstorming with no copy deliverable.

Re-run Behavior

• Re-running write.js creates fresh copy variations and may consume additional credits.
• Re-running write.js --dry-run is safe and does not call the API.
• Re-running models.js is safe and only refreshes the available chat model list.

Resources

• Ad copy patterns and channel guidance: references/domain.md