dry-three-copy-tests
Skill5 starsChanged 11 days agoApache-2.0View on GitHub
Write and revise marketing copy using Harry Dry's tests for vivid imagery, checkable claims, and earned brand specificity. Use for ads, headlines, landing pages, emails, calls to action, and copy critiques.
In this repository168 other files it gives its agents.
- abel-sell-the-alphaSkill
- agent-governance-gap-analysisSkill
- agent-infra-fixerSkill
- agent-receiptSkill
- agent-skills-gapSkill
- ai-dnaSkill
- ai-festival-posterSkill
- ai-teamSkill
- albro-founder-brandSkill
- anatomySkill
- andy-raskin-strategic-narrativeSkill
- ar-aging-and-cash-applicationSkill
- asshauer-performance-contentSkill
- autonomous-researchSkill
- awwwards-sitesSkill
- better-typographySkill
- bourgoin-why-we-buySkill
- brain-scanSkill
- brath-b2b-demand-frameworkSkill
- braun-poke-the-bearSkill
- broll-pickerSkill
- claritySkill
- client-commsSkill
- cli-ux-reviewSkill
- clone-my-voiceSkill
- clone-yourselfSkill
- costa-negotiation-dynamicsSkill
- crawler-access-auditSkill
- cv-job-matchSkill
- data-globeSkill
- decision-ledgerSkill
- denney-persona-angle-formatSkill
- dicks-homework-five-secondSkill
- domain-uptime-monitorSkill
- dreyer-funnel-diagnosisSkill
- dunford-obviously-awesomeSkill
- edge-find-skillsSkill
- fede-cloneSkill
- fede-voiceSkill
- fehr-clarity-taste-strategySkill
- fitzpatrick-mom-testSkill
- flores-founder-content-machineSkill
- founders-handbookSkill
- foxwell-meta-creativeSkill
- frankus-geo-auditsSkill
- gahan-b2b-creator-blueprintSkill
- galloway-youtube-outlier-packagingSkill
- generate-imageSkill
- geo-dominationSkill
- gerhardt-founder-brandSkill
- german-taxes-elsterSkill
- goetz-challenger-marketingSkill
- greger-creator-squadSkill
- handley-everybody-writesSkill
- harness-firstSkill
- hire-founding-engineerSkill
- hott-ugly-adsSkill
- http-error-triageSkill
- hubspot-crm-cleanupSkill
- humanizerSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 16
Before drafting, read the relevant sections of references/method.md.
- Distinctive: line 36
Make the whole unit understandable: Read the headline with its subtitle, proof, visual, and next step where those elements exist.
- Distinctive: line 38
Edit for a human reader: Read aloud and remove inflated wording, repetition, and unnecessary modifiers without stripping useful personality.
- Distinctive: line 40
Review and repair: Run checklists/review-checklist.md.
- Distinctive: line 42
Deliver the requested copy first: Use templates/output-template.md for the handoff.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Harry Dry's Three Copy Tests
Based on Harry Dry's public method. Not affiliated with or endorsed by Harry Dry.
When to use
Use this skill when marketing copy needs to help a particular reader understand an offer and take a next step. It suits both new drafts and revisions, especially polished sentences that hide the product behind abstractions or could fit almost any advertiser. The deliverable is finished copy accompanied, when useful, by a brief explanation of its imagery, evidence, and distinctive truth. [S1][S2]
Apply the tests most closely to the main selling promise. Supporting lines can explain the mechanism, substantiate the promise, and answer objections. A button or factual label need not carry an entire sales case. Adapt the scope to the requested channel and length; use this as a writing method rather than a requirement to redesign every page. [S3]
Before drafting, read the relevant sections of references/method.md. The examples illustrate decisions, not reusable claims. Source citations identify the public advice behind the method; they are never evidence that a user's product delivers a particular result.
Procedure
- Define the assignment. Record the reader, channel, desired action, length, and reader's most important problem. Use the brief fields in templates/output-template.md. Choose one primary promise so the copy has a clear job; do not silently widen a headline request into a full campaign.
- Gather usable facts. Inventory the actual offer, mechanism, constraints, proof, and objections. Separate supplied evidence from assumptions and missing information using references/method.md, section "Build the evidence base." Ask for a fact only when it changes the promise; otherwise draft within what is known.
- Listen for customer language. Extract recurring frustrations and desired outcomes from supplied interviews, reviews, or support notes. Follow "Build the evidence base" in references/method.md. Preserve meaning without inventing testimonials or making one comment represent everyone. If no customer material exists, label audience assumptions. [S4]
- Write a plain proposition. State what the reader gets and how the offer supplies it. Follow "Choose the selling angle" in references/method.md. Explain an unfamiliar product directly; for a familiar category, consider a supported objection, audience, or mechanism as the hook. [S3]
- Find the visible moment. Replace abstract benefits with relevant objects, actions, or situations. Use "Test 1: Picture" in references/method.md, and compare the tool-library illustration in examples/worked-examples.md. Keep the scene connected to the offer rather than adding decoration.
- Identify the truth test. For each important assertion, describe what observation would confirm or contradict it. Follow "Test 2: Check" in references/method.md. Replace untestable praise with a bounded fact or observable mechanism. A number without evidence remains unusable.
- Find earned specificity. Substitute a plausible competitor's name and inspect whether the line remains natural and true. Follow "Test 3: Belong" in references/method.md. Use meaningful facts about the actual offer; if no difference is established, record that limitation instead of inventing exclusivity.
- Compare concrete angles. Draft a few materially different options, such as a use moment, an objection, or a product mechanism. Apply the three tests separately using the audit fields in templates/output-template.md. Choose the strongest truthful option; a memorable but unsupported line cannot win.
- Make the whole unit understandable. Read the headline with its subtitle, proof, visual, and next step where those elements exist. Follow "Context and channel" in references/method.md. Show how the offer works and ensure the button describes a next step it actually provides. [S3]
- Edit for a human reader. Read aloud and remove inflated wording, repetition, and unnecessary modifiers without stripping useful personality. Apply "Voice and explicit warnings" in references/method.md. Retain qualifications that keep the promise accurate. [S5][S6]
- Review and repair. Run checklists/review-checklist.md. Revisit facts when the idea fails all three tests; rewrite the relevant line when a local weakness remains. Record unresolved evidence or specificity gaps explicitly. Stop when further edits add style without improving clarity, support, or relevance.
- Deliver the requested copy first. Use templates/output-template.md for the handoff. Include a concise audit and open items when the user wants reasoning. For final-copy-only requests, perform the audit internally and return only supported copy in the requested format.
Files in this skill
- SKILL.md: Entry point, usage scope, and the agent's drafting and revision loop.
- references/method.md: Detailed tests, evidence rules, channel decisions, edge cases, and warnings.
- references/sources.md: Numbered register of the six public sources actually fetched.
- examples/worked-examples.md: Invented before-and-after illustrations with rule audits.
- templates/output-template.md: Fill-in copy deliverable and field-by-field guidance.
- checklists/review-checklist.md: Pass/fail review, common failures, and targeted fixes.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

