Guide · Agent skill
How to use the Paid Ads Campaign Assistant agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The Paid Ads Campaign Assistant is an agent skill from the marketingskills collection (coreyhaines31/marketingskills, skills/ads, MIT licence). It gives an AI agent the working method of a performance marketer for campaigns on Google Ads, Meta, LinkedIn, Twitter/X and similar platforms. This guide explains what the skill does, how to install it, how to trigger it with plain requests, and what it needs before you rely on its advice.
Paid Ads Campaign Assistantcoreyhaines31/marketingskills · MIT Download the skill Quick start
- 1Download the ZIP. Get the ads skill ZIP from the directory page.
- 2Unzip into a skills folder. Unzip into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder.
- 3Or upload in the Claude apps. Go to Settings → Capabilities → Skills and upload the ZIP.
- 4Gather your data. Collect your goals, budget, targets and either account access or an export of campaign data.
- 5Ask a natural question. Describe the campaign and the decision you need, using a trigger phrase such as wasted spend or should I kill this ad.
- 6Review the answer. Check the facts, hypotheses and MISSING items, then make any change yourself after checking your own data.
What is an agent skill, and what does this one make the agent do?
An agent skill is a folder of instructions and reference files that an AI agent loads when a task matches the skill's description. You do not have to paste the instructions into each chat.
Once installed, this skill tells the agent to act like a performance marketer. Before advising, it asks about your goals, budget, audience and current results. It then uses detailed playbooks to judge whether to keep, scale, pause or test ads. It can also audit live accounts and read search-term data, and it separates facts, hypotheses and missing data in its answers.
- Campaign strategy, audience targeting, bidding and optimization across Google Ads, Meta, LinkedIn and similar platforms
- Kill, keep, scale and test decisions, backed by playbooks for Google Search, Meta, LinkedIn and B2B paid media
- Google Ads and ecommerce account audits, including wasted-spend checks on search terms
- Labelled uncertainty: unavailable inputs are marked MISSING rather than estimated
- Pointers to other skills for ad creative (ad-creative), landing pages (cro), cold email (cold-email) and tracking (conversion-tracking)
How do you install the ads skill?
Download the skill ZIP from the directory page. Then choose one of the two install routes below.
For a personal install, unzip the file into ~/.claude/skills/. For a single project, unzip it into the project's .claude/skills/ folder. Either way, the skill folder should contain SKILL.md and its references folder.
In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.
- Personal: ~/.claude/skills/
- Per project: .claude/skills/ inside your project folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
How do you trigger the skill with a natural request?
You do not need a special command. Describe your situation in plain language and mention the platform or the problem. The skill's description lists trigger phrases such as 'ROAS', 'CPA', 'wasted spend', 'negative keywords', 'search terms report', 'should I run ads' and 'is this campaign working'.
Give the agent your numbers when you have them. Targets, budget, conversion counts and spend history make the advice far more specific. If you leave something out, the skill is designed to ask or flag it as MISSING.
- 'Is this Google Ads campaign working? Spend is $1,200 this month and I have 9 leads.'
- 'Find wasted spend in this search terms export and suggest negative keywords.'
- 'Should I kill this Meta ad? Here is the last 14 days of results.'
- 'Plan a LinkedIn budget for 150 target accounts with a 90-day sales cycle.'
What do you need before you start?
For live analysis, the skill needs access to your ad platform accounts, or exported campaign data from Google Ads, Meta or LinkedIn. Without either, it can still help with strategy and planning, but it cannot judge your actual performance.
The skill also checks for a product marketing context file. If .agents/product-marketing.md or .claude/product-marketing.md exists, the agent reads its Paid Ads section for targets and floors, and its Learnings section so that it does not repeat tests you have already run.
- Access to ad accounts, or exported campaign data (Google Ads, Meta, LinkedIn)
- Your campaign goal, target CPA or ROAS, and budget
- Landing page URL and offer details
- Optional: a product marketing context file as described above
What are the safety notes?
According to its listing, the skill runs no scripts and uses no network access. Its purpose is to inform decisions on accounts the agent can reach. It is not designed to make live spending changes without human review.
The playbook thresholds are generic. Check every recommendation against your own account data before you change a budget, bid or ad.
- Runs scripts: no
- Uses network: no
- Review every change yourself before applying it in an ad platform
- Treat thresholds as starting points, not guarantees
What tips help you get better answers?
Ask for the decision you actually face, such as keep, scale, pause or test, and include the period and the data you have. Ask the agent to list facts, hypotheses and MISSING inputs separately, since the skill is built to do this. For B2B and ABM questions, say how long your sales cycle is and how big your target account list is, because the playbooks use those numbers.
- Give the time window and the conversion definition you use
- Name the platform, since each has its own playbook
- Ask for a short weekly readout when you want a stakeholder summary
- For tracking or pixel problems, ask for conversion-tracking instead of paid ads advice
What people use it for
- Deciding whether to keep, scale, pause or test a Google Ads or Meta campaign from recent results
- Finding wasted spend in a Google Ads search terms report and drafting negative keywords
- Auditing a small Google Ads or Shopping account against a structured checklist
- Planning LinkedIn or account-based ad budgets for a named target list
- Writing a short weekly paid media readout for stakeholders
Frequently asked questions
How do I install the ads agent skill?
Download the ZIP from the directory page and unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. In the Claude apps, you can upload the ZIP under Settings → Capabilities → Skills instead.
Is the ads skill free to use?
The skill is published under the MIT licence, as listed on the directory page. Check the licence file in the repository for the full terms.
What do I need to use it for live account analysis?
You need access to your ad platform accounts or exported campaign data from Google Ads, Meta or LinkedIn. Without this, the skill can still help with strategy and planning but cannot judge actual performance.
Does the ads skill run scripts or access the internet?
According to its listing, it runs no scripts and uses no network access. It informs decisions rather than making live spending changes, so you should review any change before applying it.
How is this different from ad-creative or cro?
The ads skill covers campaign strategy, targeting, bidding and optimization. Bulk ad creative generation belongs to the ad-creative skill, landing page work to cro, and pixels and tracking audits to conversion-tracking.
Why does the skill say MISSING instead of giving a number?
It is designed to label unavailable inputs as MISSING rather than estimate them. Supply the missing data, such as conversion counts or spend history, and the advice will be more specific.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you run paid campaigns and want structured keep, scale, pause or test decisions, or if you are a small team auditing a Google Ads or Shopping account for wasted spend. Pick something else if you expect guaranteed results or want the agent to change live spend without your review, since its thresholds are generic and must be checked against your own data.
Download the skill