Systems Lab

Agent skill

ads

Build ad campaign plans grounded in your Octave library.

activeReaches the webActs undeclared359 words

Filed under ABM and paid.

From octavehq/lfgtm · 27 skill entries · 11 · pushed 2026-09-21

What it does when it runs

Build ad campaign plans grounded in your Octave library. Generates ad sets with platform-ready creative, audience targeting, negative keywords, and landing page recommendations, every variant derived from an auditable source card built from real prospect language and proof points. Use when user says \"build an ad campaign\", \"create ads\", \"ad campaign for\", \"generate ad sets\", or asks for paid advertising creative. Do NOT use to analyze ad performance: use /octave:ads-resonance (the resonance loop) instead.

Automated analysis of the skill and the 7 files bundled beside it. A skill’s own description is written to be selected by an agent, so it describes the job and not the dependencies.

Keys and connectors you must supply
None found.
Hosts it reaches
  • developers.google.com
  • support.google.com
Tool permissions it declares
No allowed-tools in the frontmatter. It does act, so it runs under whatever permissions your session already grants.
Actions present in the files
shellwrites files

Ask about ads

Opens your assistant with this page's verified links already in the prompt.

Is this safe to install?ClaudeChatGPT
Adapt it to my stackClaudeChatGPT
What else do I need for it to workClaudeChatGPT
Rather ask a human? Talk to Cheetah

Install it

View source on GitHub ↗
git clone --depth 1 --filter=blob:none --sparse https://github.com/octavehq/lfgtm.git /tmp/lfgtm
git -C /tmp/lfgtm sparse-checkout set "skills/ads"
mkdir -p ~/.claude/skills/ads
cp -R "/tmp/lfgtm/skills/ads/." ~/.claude/skills/ads/

Picked up without a restart. A project skill of the same name is shadowed by your personal one. For one repository only, swap ~/.claude/skills for .claude/skills. Claude Code docs ↗

Or take the whole library

This repo ships a .claude-plugin manifest, so Claude Code can install all 27 skills at once. Plugin skills are invoked as /<plugin>:<skill>, so they never collide with your own.

/plugin marketplace add octavehq/lfgtm
/plugin

The folder is the same in every client that implements the format — 46 of them — so if yours is not above, only the destination changes.

The skill

Source on GitHub ↗

Reproduced in full from octavehq/lfgtm/blob/6facb2c00d93c48e29a370d3d93d52e4380f3b45/skills/ads/SKILL.md, which is licensed MIT (repository). 359 words, 5 headings.

Octave Ads — Campaign Generation

Run the task

Read the task method before selecting context, claims, or output. It is the execution method for this skill. Supporting templates supply structure and styling; populate them from the task method and current evidence, not their illustrative figures or assertions.

Resolve platform, objective, child advertising account, offer and CTA. Preserve paused defaults for exports. Google RSA exports require 3–15 headlines, 2–4 descriptions and final URL; five headlines is a recommendation. Meta/LinkedIn are copy-and-creative handoffs until their import schemas and media packages are verified. No-persist means no source-card file reads or writes.

Use GTM context for strategic tasks and evidence and inputs for material claims and missing facts. A narrow list/get or a render-only handoff does not require unrelated research. Read host runtime for available tools, portable resource paths, routing, and review fallback.

Resolve current tool schemas before execution; use returned IDs and pagination. Inventory rows locate records; hydrate selected entities/resources before relying on their contents. Reuse supplied or inherited context and authorization. Ask grouped questions only for material unresolved inputs, continue independent work, incorporate replies, and recheck affected output.

Usage

Use /octave:ads with the requested task, target, and output.

References by task

Read only the reference for the selected mode, then the format/layout references when rendering. Do not repeat intake or change approved claims when routing to a renderer.

Delivery and changes

Apply output readiness to text and artifacts. HTML also uses the review protocol and the matching format checks. A working preview can be shared with its status; unrun checks are NOT RUN. Reviewers return findings on an immutable version; one author applies edits and rechecks the final version.

For visual output, inherit the selected sender/workspace brand through brand kit usage; honor an explicit override and avoid repeated brand extraction. Public hosting requires the approved audience and publish manifest, followed by URL/access/link verification.

For comparisons, trends or rates, use the analytics contract. Only requested persistence loads workspace mutations; report changes as applied only after scoped readback.

Files bundled with it

These load only when the skill asks for them, so they cost nothing until it runs.

Other skills for the same job

Different authors, same problem. Matched on the words in the skill name, across every library in the catalogue except this one.

Need help setting it up?

This page tells you what ads does and what it needs. Cheetah builds the agent setup it runs inside: data, CRM, sequencing and the guardrails.

Book a call →

The directory stays free. There is nothing gated behind this.