What does this skill do, and when should you use it?
Subject Line Lab focuses on email subject lines and preheaders, not body copy, CTAs, test design, or EQS auditing. It generates 3–8 labeled variants from an offer or angle, or scores candidates supplied by the user. Each variant receives heuristic checks for spam-trigger patterns, desktop and mobile truncation, emoji count, and the rendered inbox preview. It is designed to cut weak candidates before testing, but its pre-score is not an inbox-placement or deliverability verdict.
Reads subject candidates or an offer/angle, from-name, email mode, preheader or drafting intent, and optional past-campaign exports. It generates or ingests 3–8 variants, assigns stable IDs such as SUBJ-A and SUBJ-B, and pairs each subject with a preheader. It measures subject and preheader character counts, compares them with estimated desktop and mobile render limits, and flags truncation when the load-bearing promise falls beyond the mobile cut. It scans subjects and preheaders for all-caps runs, repeated exclamation marks, misleading RE/FWD fakery, false scarcity, spam-word density, and stacked currency or percent symbols; counts emojis; renders the from-name, subject, and preheader preview; labels findings as Measured or Estimated; and produces a ranked survivor/cut list plus a handoff summary.
- An ecommerce marketer needs several promotional or lifecycle subject/preheader variants before launching a campaign.
- A B2B sales team wants to pre-screen cold-outbound subjects for truncation, spam-pattern flags, and emoji usage.
- A newsletter operator has candidate subjects and wants to rank them before creating an A/B test.
- An email team has historical subject-line and open-rate exports and wants to use prior winning angles in the shortlist.
How do you install this skill?
- Do not treat the heuristic pre-score as a Gmail/Outlook placement or spam verdict; real rendering, authentication, and deliverability checks remain necessary.
- Key references such as subject-line-specs.md, humanizer-slop.md, and skill-contract.md were not supplied, so all thresholds, templates, and write protocols cannot be confirmed as reproducible.
- In cold-outbound mode avoid emojis and verify every price, discount, scarcity, and personalization claim before use.
- Shell / CLI
- Local filesystem
registry-events.py when submitting claim proposals
Install the repository with npx skills add aaron-he-zhu/aaron-marketing-skills. In Claude Code, use /plugin marketplace add aaron-he-zhu/aaron-marketing-skills, then /plugin install aaron-marketing@aaron. The repository can also be cloned with git clone https://github.com/aaron-he-zhu/aaron-marketing-skills; this skill is at email/engage/subject-line-lab/SKILL.md. The source gives a single-skill installation pattern using npx skills add <repo> -s <skill>, but does not provide this skill's exact command.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/aaron-he-zhu/aaron-marketing-skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/email/engage/subject-line-lab" ~/.claude/skills/
rm -rf "$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?
Provide subject candidates or an offer/angle, a from-name, a mode (promo, cold, or newsletter), and a preheader or permission to draft one. Example: Pre-score these 6 subject lines for truncation + spam triggers, from-name [Sender], promo mode: [paste]. Another supported prompt is Generate 5 subject-line variants + preheaders for [offer], cold-outbound mode, and rank them by pre-score. If neither candidates nor an offer/angle is supplied, the skill returns NEEDS_INPUT. Missing from-name or preheader can be handled with a placeholder or an estimated drafted preheader.
What are this skill's strengths and limitations?
- Covers four practical pre-test checks: spam patterns, desktop/mobile truncation, emoji count, and inbox preview.
- Uses stable variant IDs and angle labels that can flow into send-experiment-designer.
- Works with pasted data and local character counts without requiring a paid ESP API or MCP.
- Separates Measured from Estimated evidence and explicitly avoids presenting the heuristic as an EQS or placement verdict.
- Render limits and spam checks are heuristics, not provider-level filtering or client-render tests.
- It does not write body copy or CTAs, design A/B tests, or calculate EQS.
- The source provides no specific ESP, mailbox-client, or live inbox-test results for this skill.
- It will not invent prices, discounts, scarcity, or other unsupported claims; unresolved claims are marked `[needs source]`.
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 |
|---|---|---|---|---|
| Subject Line Lab this page | 56 · Use with care | ★ 3k | 12d ago | Apache-2.0 |
| Inbox Placement Monitor | 59 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Email Creative Builder | 58 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Email Deliverability Preflight | 61 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Email Marketing Bible | 60 · Recommended | ★ 342 | 14d ago | MIT |
Its role is distinct from the repository's named neighboring skills: email-creative-builder writes the email body and CTA, send-experiment-designer designs and reads tests, deliverability-qa performs broader content and authentication checks, and email-quality-auditor computes SEND/EQS. Subject Line Lab only pre-screens the subject/preheader unit.
How did FollowSkills review this skill?
The skill treats exports and pasted content as untrusted, avoids requiring paid APIs, asks before writing memory, and submits claims through a proposal path. No covert exfiltration, credential theft, or destructive default is shown. Points are deducted because rollback and recovery for memory writes are not specified in the skill itself, referenced registry controls are not fully included in the supplied evidence, and publisher identity is unverified.
Inputs, outputs, workflow steps, stop conditions, and Measured/Estimated labeling are internally coherent, with a NEEDS_INPUT path for missing or ambiguous inputs. Points are deducted because key referenced files were not supplied and there is no skill-specific test suite or execution evidence; static calibration caps this at 10.
The audience, three operating modes, use cases, non-fit boundaries, trigger phrases, and output shape are clearly stated, and core computation is described as local rather than dependent on overseas services. Points are deducted because Chinese support is mostly metadata and labels with no Chinese examples, and cross-client rendering limits remain heuristic estimates.
The skill provides a name, slug, version 18.0.0, compatibility, Apache-2.0 license, Quick Starts, contract, data sources, decision gates, and routing guidance with good layering. Points are deducted because maintenance ownership, update procedure, changelog, and complete installation guidance are outside the target skill; the supplied SECURITY.md calls 17.x the current line, conflicting with version 18.0.0 governance.
It specifies a directly usable deliverable: 3-8 stable-ID subject/preheader variants with length, truncation, spam-pattern, emoji, preview, ranking, and handoff information. Points are deducted because scoring is explicitly heuristic, actual client rendering and filtering accuracy are unverified, comparative value over manual review is unproven, and static review provides no execution evidence; therefore this stays at or below 7.
The files distinguish Measured character counts from Estimated rendering and spam heuristics and disclose important limitations and downstream ownership. Points are deducted because there are no real outputs, skill-specific tests, or independent third-party corroboration; the supplied CI tests concern architecture and audit validation rather than this skill's key paths.
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 →