Skip to content
skillsafebeta
← Report library
Purpose / Writing

Copy Editing Skill Security Audit

What the author says it does (original text)

When the user wants to edit, review, or improve existing marketing copy, or refresh outdated content. Also use when the user mentions 'edit this copy,' 'review my copy,' 'copy feedback,' 'proofread,' 'polish this,' 'make this better,' 'copy sweep,' 'tighten this up,' 'this reads awkwardly,' 'clean up this text,' 'too wordy,' 'sharpen the messaging,' 'refresh this content,' 'update this page,' 'thi

Independent security check

Security risks found

Files checked
5
Risks found
3
Could it run dangerous commands?Looks for programs run straight after downloading, remote control of your computer, and hidden commands.No risks found
Could it expose your files or keys?Looks for uploads of files containing passwords or keys, and keys written directly in the code.Risks found: 1
Medium risk

Automatically reads additional internal marketing context

Source references: 1
What we found

Even when the user only asks to edit supplied copy, the Skill directs the agent to first read a product-marketing file from one of three preset locations. That file could contain customer research, positioning, or other internal material outside the request.

Why this matters

The material enters the agent's working context, expanding the data exposed to the task. The supplied evidence does not instruct external transmission, but the user may not intend a copy-editing request to access the full internal document.

If one of the predefined paths exists, the agent is explicitly told to read it before editing. Although relevant to brand voice, that file may contain customer research, positioning, or other internal information beyond the copy the user submitted. A user can require the Skill to use only supplied text, or to identify the file and obtain consent before reading it.

SKILL.md:14In the instructionsOpen original file
**Check for product marketing context first:**If `.agents/product-marketing.md` exists (or `.claude/product-marketing.md`, or the legacy `product-marketing-context.md` filename, in older setups), read it before editing. Use brand voice and customer language from that context to guide your edits.
Could it delete files or keep running?Looks for broad file deletion, disk overwrites, and programs set to start automatically.No risks found
Could it bypass safety checks?Looks for skipped website security checks, excessive file access, or actions that skip your approval.No risks found
Could it mislead the AI or hide text?Checks the skill instructions for requests to ignore you, influence the report, or hide text in invisible characters.Risks found: 1
Medium risk

A simulated “expert panel” may be mistaken for independent professional review

Source references: 3
What we found

For high-stakes copy, the Skill directs the agent to assemble 3–5 “expert personas,” score the copy, and iterate to numerical thresholds. It does not require disclosure that these are personas simulated by one agent rather than real, independent experts or customers.

Why this matters

A user could treat an average score of 8+ as genuine validation by multiple experts, customers, or compliance reviewers and publish launch, pricing, or sales copy on that basis. Persona simulation cannot verify facts, legal compliance, or actual customer response.

The workflow explicitly assembles expert personas and has them score the copy rather than consulting real experts; one suggested reviewer is a target-customer persona. It presents this simulation as addressing the limits of a single perspective and uses score thresholds, without requiring disclosure that one model is role-playing every reviewer. Users could mistake this for independent expert or customer validation. They can require it to be labeled a simulated heuristic review, not customer research or independent professional review.

SKILL.md:259In the instructionsOpen original file
## Expert Panel ScoringUse this after completing the Seven Sweeps for an additional quality gate. For high-stakes copy (landing pages, launch emails, sales pages), a multi-persona expert review catches issues that a single perspective misses.### How It Works1. **Assemble 3-5 expert personas** relevant to the copy type2. **Each persona scores the copy 1-10** on their area of expertise3. **Collect specific critiques** — not just scores, but what to fix4. **Revise based on feedback** — address the lowest-scoring areas first5. **Re-score after revisions** — iterate until all personas score 7+, with an average of 8+ across the panel
Show 2 other places
SKILL.md:301In the instructionsOpen original file
### When to Use- **Always** for launch copy, pricing pages, and high-traffic landing pages- **Recommended** for email sequences, sales pages, and ad copy- **Optional** for blog posts, social content, and internal docs- **Skip** for quick updates, minor edits, and low-stakes content
SKILL.md:273In the instructionsOpen original file
**Landing page copy:**- Conversion copywriter (clarity, CTA strength, benefit hierarchy)- UX writer (scannability, cognitive load, user flow)- Target customer persona (does this speak to me? do I trust it?)- Brand strategist (voice consistency, positioning accuracy)
Could it change links or payment recipients without asking?Looks for forced referral or payment changes combined with instructions to hide the change.Risks found: 1
Medium risk

May add unconfirmed guarantees, trials, or cancellation promises

Source references: 5
What we found

The “Zero Risk” pass tells the agent to add risk reversals or trust signals as needed and lists money-back guarantees, free trials, “no credit card required,” and “cancel anytime.” It does not require confirmation that the business actually offers or authorizes those terms.

Why this matters

If inserted directly into pricing, sales, or CTA copy, customers may rely on nonexistent refund, trial, cancellation, or privacy promises, creating complaint, refund, regulatory, or contractual exposure.

This is an active editing workflow: it lists refunds, trials, no-card signup, and cancellation as real commercial terms, then says to add them as needed. An earlier sweep calls for proving or softening claims, but it does not expressly prohibit inserting these promises before verifying the merchant's policies. If suggestions are placed directly into publishable copy, users could advertise terms that do not exist. Users can require such language to remain labeled placeholders unless confirmed by the business.

SKILL.md:223In the instructionsOpen original file
### Sweep 7: Zero Risk**Focus:** Have we removed every barrier to action?**What to check:**- Friction near CTAs- Unanswered objections- Missing trust signals- Unclear next steps- Hidden costs or surprises**Risk reducers to look for:**- Money-back guarantees- Free trials- "No credit card required"- "Cancel anytime"- Social proof near CTA- Clear expectations of what happens next- Privacy assurances
Show 4 other places
SKILL.md:249In the instructionsOpen original file
**Process:**1. Focus on sections near CTAs2. List every reason someone might hesitate3. Check if the copy addresses each concern4. Add risk reversals or trust signals as needed
references/checklist.md:54In the instructionsOpen original file
## Zero Risk (Sweep 7)- [ ] Objections addressed near CTA- [ ] Trust signals present- [ ] Next steps are crystal clear- [ ] Risk reversals stated (guarantee, trial, etc.)
SKILL.md:142In the instructionsOpen original file
**Process:**1. Identify every claim that needs proof2. Check if proof exists nearby3. Flag unsupported assertions4. Recommend adding proof or softening claims
SKILL.md:234In the instructionsOpen original file
**Risk reducers to look for:**- Money-back guarantees- Free trials- "No credit card required"- "Cancel anytime"- Social proof near CTA- Clear expectations of what happens next- Privacy assurances

Inside this skill

8 instruction sections

The Skill edits existing marketing copy through seven passes covering clarity, voice, customer benefit, proof, specificity, emotion, and barriers to action, with earlier checks revisited after changes.

View source
SKILL.md:27In the instructionsOpen original file
## The Seven Sweeps FrameworkEdit copy through seven sequential passes, each focusing on one dimension. After each sweep, loop back to check previous sweeps aren't compromised.

Before editing, it looks for a product-marketing context file in the workspace and uses its brand voice and customer language to guide edits.

View source
SKILL.md:14In the instructionsOpen original file
**Check for product marketing context first:**If `.agents/product-marketing.md` exists (or `.claude/product-marketing.md`, or the legacy `product-marketing-context.md` filename, in older setups), read it before editing. Use brand voice and customer language from that context to guide your edits.

For published content, it recommends verifying facts, pricing, features, and links and refreshing pages on a schedule; the supplied files contain no scripts, installation steps, or network-transmission instructions.

View source
references/content-refresh.md:15In the instructionsOpen original file
1. **Freshness pass** — Update all dates, stats, and examples. Replace "in 2024" with current data. Remove references to deprecated features or tools.2. **Accuracy pass** — Verify all claims are still true. Check that linked resources still exist. Confirm pricing and feature descriptions match current state.3. **Voice pass** — Does the tone match your current brand voice? Older content often reflects an earlier stage of the company.4. **SEO pass** — Has search intent shifted for this topic? Are there new keywords or questions to address? Add "Last updated: [date]" prominently.5. **Proof pass** — Can you add newer testimonials, case studies, or data points that didn't exist when this was first published?6. **Structure pass** — Add comparison tables, FAQ sections, or other scannable formats that make the content easier to consume.
references/content-refresh.md:32In the instructionsOpen original file
## Refresh Cadence- **Pricing and product pages**: Every quarter, or when pricing/features change- **High-traffic blog posts**: Every 6 months- **Comparison and alternatives pages**: Every 3-6 months (competitors change fast)- **Evergreen guides**: Annually, unless traffic drops sooner- **Low-traffic pages**: Only when traffic data suggests an opportunity
Start here · InstructionsSKILL.md
copy-editing
Lines connect the instruction file to its sections, not an observed execution order. Select a section to read the source. 4 more sections are available in the original file.

File reference map

References: 3
Files making referencesReferenced content
Lines show actual file references, not execution order. Select a node to highlight its connections and inspect the files and source locations. Dashed lines include files that still need locating.
Files and check records5 files

Coverage and gaps

Content covered in each file

These are the source ranges included in this check, not a guarantee that every issue has been resolved.

  • SKILL.mdFull text included
  • references/checklist.mdFull text included
  • references/content-refresh.mdFull text included
  • references/plain-english-alternatives.mdFull text included
  • evals/evals.jsonFull text included

This report is for the version above. We read the available code and instructions without running the skill or checking extra packages it installs. This is not a promise of safety: a different version or setup may behave differently.

  • SKILL.mdInstructions
  • evals/evals.jsonSupporting file
  • references/checklist.mdSupporting file
  • references/content-refresh.mdSupporting file
  • references/plain-english-alternatives.mdSupporting file
Lines read
1,049
File checksum (to compare versions)
c90b7b01c5089cc87e4c45ebcc695d0736e06946ee0f8352a79430c1d99f8122