What does this skill do, and when should you use it?
Message Test Designer creates validation designs for candidate brand messages; it does not run experiments or analyze results. It covers comprehension, five-second recall, and Wynter-style message-market-fit protocols. The resulting specification defines the hypothesis, panel and recruiting criteria, stimulus set, thresholds, estimated panel size, and stop-or-revise rule. It fits teams that already have a candidate tagline, positioning statement, or surface-specific message.
Reads the message house, narrative canon, candidate variants, and approved claim wording; confirms the message and decision under test; converts “does it land?” into a measurable hypothesis; selects a comprehension, five-second, or message-market-fit protocol; assembles stimuli from the canon and marks unapproved claims with “[needs source]”; defines panel criteria, estimated sample size, pass thresholds, stop/revise rules, and the execution handoff; and may calculate significance from returned proportion counts with a Python script, without operating the panel.
- A brand team has a new tagline and wants to validate comprehension before increasing paid spend.
- A product marketer needs to compare several positioning statements with a target-role panel.
- A web team wants to test whether visitors recall the core message from a new homepage hero after five seconds.
- A narrative lead needs a Wynter-style message-market-fit test design for candidate messaging.
How do you install this skill?
- This skill designs tests but does not run panels or adjudicate results; separately confirm platform access, panel quality, data privacy, and mainland-China reachability.
- It depends on the message house, narrative registry, and claims ledger; missing inputs should cause a stop rather than authoring or approving the message.
- The supplied files do not independently substantiate the cited Wynter guidance or statistical-tool claims; sample sizes and thresholds remain hypotheses to validate.
- Shell / CLI
- Local filesystem
Python 3
Install the repository collection with Claude Code by running /plugin marketplace add aaron-he-zhu/aaron-marketing-skills, followed by /plugin install aaron-marketing@aaron. On other Agent Skills-compatible hosts, run npx skills add aaron-he-zhu/aaron-marketing-skills, or clone it with git clone https://github.com/aaron-he-zhu/aaron-marketing-skills. The source does not document a standalone installation command for this individual skill.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/aaron-he-zhu/aaron-marketing-skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/narrative/evaluate/message-test-designer" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- [headline + subhead]
Provide the candidate message, target panel, and variants. Example: Design a 5-second comprehension test for our new homepage hero: "[headline + subhead]". What do we measure and what's the pass bar? The skill produces a test design specification; execution is handed to send-experiment-designer or ad-test-designer, and result analysis to performance-analyzer.
What are this skill's strengths and limitations?
- Produces a concrete specification covering hypothesis, panel, protocol, thresholds, and failure handling.
- Requires stimuli to come from the existing message canon and flags unapproved claims.
- Can design the test without a paid tool.
- Clearly separates design, execution, analysis, and claim adjudication.
- Does not recruit a panel, run an experiment, or fully analyze returned results.
- Requires a candidate message, message canon, and claims context; without a candidate message it stops for input.
- Panel-size guidance is Estimated and assumption-based, not measured evidence.
- The source provides no standalone test suite or platform execution evidence.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Message Test Designer this page | 61 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Category Narrative Map | 58 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Narrative Quality Gate | 60 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Narrative Cascade Planner | 58 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Audience Belief Mapper | 57 · Use with care | ★ 3k | 12d ago | Apache-2.0 |
How did FollowSkills review this skill?
The skill clearly limits itself to test design, never executes experiments, treats pasted material as untrusted, marks unsupported claims as [needs source], and requires confirmation or authorized proposal flows for memory and registry writes. Six points are deducted because panel-data minimization, sensitive-data handling, recovery, and external-platform boundaries are not specified in detail.
Triggers, the NEEDS_INPUT branch, protocol selection, thresholds, failure revision path, and handoff boundaries are internally clear. Eleven points are deducted because this is a static review with no skill-specific tests or reproduced key paths, and external panel execution plus referenced-resource availability are unverified.
The audience, message inputs, three protocol choices, non-fit boundaries, parameters, and Chinese/English triggers are reasonably clear, with a keyless design path. Four points are deducted because execution depends on external platforms whose mainland-China reachability and substitutes are not addressed, and the target-panel boundary still requires user input.
The file provides structured frontmatter, version 18.0.0, Apache-2.0 licensing, Quick Start prompts, contract, data sources, limitations, handoff, and next-skill guidance. Four points are deducted because there is no concrete output template, FAQ, or troubleshooting section, maintenance ownership and update path are unclear, and publisher identity is unverified.
The expected spec covers a measurable hypothesis, recruit criteria, stimulus set, thresholds, estimated sample-size assumptions, stop/revise rules, and execution handoff, so the core design task is well covered. Eight points are deducted because no representative output or execution evidence is included, and the result depends on existing message-house, claims-ledger, and downstream builder context.
The skill references repository-local contract, TALE, security, and connector materials and requires separating Measured, User-provided, and Estimated data. Six points are deducted because the cited Wynter guidance, threshold examples, and effectiveness claims lack verifiable third-party evidence in the supplied files, and no skill-specific reproducible tests are provided.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →