Growth Agent Starter Kit
Three complete marketing agents with a worked example of each, a run log, and a scorecard that tells you within five runs whether to keep going. Runs in a chat window or a coding agent.
Free. View purchase detailsBuilt for: A marketer or founder who wants to run one agent on real work, with a human gate, before deciding whether agents belong in their marketing at all.
Your first deliverable
a suggested thirty-minute checklist to hand a page owner one reviewed landing page and a named change.
What the Offer Page Reviewer catches
Worked example in the kit · fictional business
The included review uses an invented payroll business to show how the agent challenges an unsupported promise.
- Input
- A landing page promises a faster pay run.
- Finding
- Testimonials, customer count, and setup time do not measure pay-run time.
- One change
- Measure a real pay run before putting a time claim beside the headline.
Based on examples/offer-page-review-worked.md. Beacon Payroll and its numbers are fictional.
Included in the workflow
- Offer Page Reviewer
- Campaign Readiness Auditor
- Content Repurposing Operator
- Three worked examples
- Thirty-minute quickstart
- Run log template
- Scorecard
- Coding-agent skills
Check the fit before buying
A marketer or founder who wants to run one agent on real work, with a human gate, before deciding whether agents belong in their marketing at all.
- An AI assistant that can read the supplied files
- A page or campaign draft to review
Read the package documentation
What follows is the README from inside the zip, unedited. The page and the package say the same thing on purpose.
Free. Version 1.3.0. Licensed to you; see LICENSE.md.
The fastest way to learn what an agent should and should not do in your marketing is to run one small one, with a human gate on it, on real work. This kit is that first loop: three complete agents, a worked example of each, a scorecard, and the log that tells you within five runs whether to keep going.
Start here
START-HERE.md offers a ten-minute entry exercise. QUICKSTART.md is a suggested thirty-minute checklist from unzipped folder to one reviewed page and one logged run. It picks the offer page reviewer for you, because that agent needs nothing you do not already have: a page, a sentence about who it is for, and the action you want.
If you would rather be interviewed, paste SETUP.md into your agent. It asks four questions, picks the one loop that fits what is in front of you, shows you the worked example, and runs the first pass with you.
Either way, read the worked example for your agent before the first run. Reading examples/ shows what a complete output looks like and what an invented number looks like.
Two ways to run it
In a chat window. Open the agent file under agents/, copy the prompt starter at the bottom, and paste it with your inputs. To make the instructions persist between runs, set up a Claude Project or a custom GPT with projects/growth-agent-starter-kit.project.md. No terminal.
In a coding agent. Open the unzipped folder in the agent you already use and ask it to read the relevant skills/<name>/SKILL.md plus the matching example. No global skill installation or account-setting change is required.
For an optional portable copy into a new folder you explicitly choose:
bash install.sh /absolute/path/to/a/new/starter-kit-folder
bash doctor.shThe installer copies only files listed in the package manifest. It refuses an existing destination, copies no unrelated working files and creates no links in home or runtime settings folders. The doctor checks the local package without reading environment files or credentials. See INSTALL.md for removal and the limits of its result.
Both forms carry the same procedure, output contract, and guardrails. Choose by where you already work, not by which sounds more serious.
What is inside
| File | What it is |
|---|---|
QUICKSTART.md | Thirty minutes, with a time budget per step, from unzip to one reviewed artifact and one logged run |
SETUP.md | The interview version. Paste it into your agent; it picks one loop and runs it with you |
README.md | This file |
scorecard.md | The primary question, the keep-or-drop bands at five runs, the trust ratio, and the guardrails that outrank all of it |
agents/offer-page-reviewer.md | Names the gap between what a landing page promises and what it proves, counts the competing actions, and returns exactly one change |
agents/campaign-readiness-auditor.md | Reviews a campaign before launch. Separates copy problems from operational blockers, treats unsourced claims as blockers, and returns a verdict |
agents/content-repurposing-operator.md | Rebuilds one source for each destination rather than trimming it. The claim stays fixed; the shape changes |
examples/offer-page-review-worked.md | A complete review of an invented company's page, in the output contract, with notes on what to check in yours |
examples/campaign-readiness-worked.md | A complete Not ready report on an invented launch, with notes on why each finding sits on the list it does |
examples/content-repurposing-worked.md | One invented article rebuilt for three destinations, including a claim the drafts added and the step that caught it |
templates/run-log.md | The run log as a fill-in table, two example rows, and the verdict column that feeds the trust ratio |
templates/agent-build-brief.md | Turns a messy idea into a scoped operator with inputs, an approval gate, and a run log |
tools/marketing-system-map.md | Decide which workflow an agent should help with first, before you build anything |
tools/launch-gate-checklist.md | The approvals and evidence a change needs before it touches a live system |
skills/offer-page-reviewer/SKILL.md | The coding-agent form of the offer page reviewer |
skills/campaign-readiness-auditor/SKILL.md | The coding-agent form of the campaign readiness auditor |
skills/content-repurposing-operator/SKILL.md | The coding-agent form of the content repurposing operator |
projects/growth-agent-starter-kit.project.md | Instructions and upload list for a Claude Project or a custom GPT |
install.sh | Copies this portable kit to one explicit new destination; refuses overwrite |
doctor.sh | Checks required files; missing files return a failure |
CHANGELOG.md | What changed in each version |
LICENSE.md | The licence |
Each agent carries a procedure, an output contract, a measurement, and guardrails. They are operators, not prompts: the procedure is ordered on purpose, and the contract stops the output drifting into prose.
The one thing most people skip
Keep the run log. One row per finding: what the agent said, what was actually true, what you did, and a verdict of correct, wrong, or unclear.
Without it, in a month you will not know whether this helped, and you will drop it for the wrong reason. With it, you will know within five runs. The table is in templates/run-log.md, with two rows filled in so you can see the grain. The bands are in scorecard.md.
The rules
- Draft and review before any send, publish, purchase, or settings change.
- Customer data and credentials stay out of prompts unless the system has explicit permission and proper storage.
- Connected is not the same as working. A tool connection is not proof the workflow produces correct outcomes. Only a verified output is.
- No number ships without a source. Not softened, not rounded. Removed.
What this kit is not
It is not a demo, and nothing here is deliberately limited to make you buy something. These three agents are complete and they are yours.
What it is not is a system. There is no orchestration, no schedule, no shared context between the three, and no scorecard rolling up across them. You run each one by hand and you keep the log yourself.
You can keep using these three complete agents without buying anything. Paid file products add defined analysis tools and report workflows; they do not automatically include hosted schedules, personal setup or a managed service. Check the actual product page for its included capabilities.
Resource delivery and optional follow-up email depend on your request and consent. This download does not subscribe you, start a sequence or send any message.
Before you start
The files
Offer Page Reviewer. Campaign Readiness Auditor. Content Repurposing Operator. Three worked examples. Thirty-minute quickstart. Run log template. Scorecard. Coding-agent skills.
Setup requirements
An AI assistant that can read the supplied files A page or campaign draft to review
The first task
Thirty minutes from unzip to one reviewed landing page with a single named change and your first logged findings.
Who runs the work
You install and run the files, inspect their output and decide what to use. A file purchase does not include consulting, setup calls or working sessions.
Payment and updates
These files are free. Marketing emails are a separate, optional choice.
Also in Free starters
Put the workflow to work.
Read the requirements, bring your inputs, and review the result before you use it.