Tier-2 Review: email-marketing (content-writing cluster)

What we attempted

We pulled the email-marketing skill from the skillsmp.com listing, extracted the SKILL.md, and ran it through the standard two-stage harness: a dependency/install probe and a smoke invocation. The intent was to confirm the skill could be stood up and exercised end-to-end before assigning a confidence rating to its composite score of 4.1/5.0.

What failed

Both harness stages were skipped, and the run closed at 0 tests passed, 0 partial, 0 failed. This is not a pass — it is an absence of signal, and we are recording it as such.

The root cause is specific and worth naming precisely: the SKILL.md is a static description of an HTML email layout, not an executable or invocable skill artifact. There is no install command, no package manifest, no dependency declaration, and no README describing an entrypoint. The truncated content we received confirms this — it is a prose spec for a "brand product-launch email" with a masthead, hero image block, headline lockup, body copy, primary CTA, and specifications grid, rendered as "pure HTML email layout (centered single column, table fallback)."

• install (skipped): no install command, manifest, or dependency block exists to execute.
• smoke-invocation (skipped): no runnable entrypoint or invocation interface is described.

In other words, the harness had nothing to invoke. The skill is a specification of an output, and our smoke test is built to invoke a capability. Those two things are not the same, and the mismatch is the failure mode — not a crash, not a wrong result, but an unbridgeable gap between what the artifact is and what the harness expects.

What we observed

The auto-summary rates this as a "high quality skill with clear triggers and output," and the dimension scores back that up on the surface: D1 trigger clarity (4.5) and D2 output specificity (4.5) are both strong. The trigger language is genuinely good — it names "email", "newsletter blast", "MJML", "product launch email", and "email template" as activation phrases, which is exactly the kind of concrete, synonym-rich trigger set that reduces misfires.

But the lower dimensions tell the real story. D4 self-containment (3.5) and D5 reusability (3.5) are the weakest scores, and our harness run explains why: a skill with no invocation surface is only reusable by a human or agent that reads the prose and reproduces the layout by hand. There is no contract, no parameters, no callable boundary. That is a legitimate design for a template, but it caps how much the harness — or any automated consumer — can verify.

Rating caveat

The 4.1/5.0 composite is therefore theoretical. It reflects the quality of the written specification, not a validated runtime behavior. Until a physical re-run resolves the harness gap — either by adding an invocation interface the harness can call, or by reclassifying this skill as a non-executable template and testing it against a render/inspection harness instead — the score should be read as a document-quality estimate, not a verified capability rating. We are not marking this as a pass, and we are not marking it as a fail. We are marking it as unverified.

Is the skill still valuable in principle?

Yes, with a caveat about what kind of value it delivers. As a design/template skill, the artifact is coherent: it specifies a complete, recognizable email layout with a clear structural inventory (masthead, hero, headline lockup, CTA, spec grid) and a sensible technical constraint (single-column, table fallback for client compatibility). For a human designer or an agent that ingests prose specs, that is directly useful and the trigger clarity makes it easy to route to.

The limitation is that it is a template, not a tool. It will never satisfy an install-and-invoke harness, and that is a category mismatch rather than a defect. The honest path forward is to test it as what it is — render the described layout and inspect the output — rather than to force it through an invocation harness it was never built to satisfy. Do that, and the theoretical 4.1 can become a measured one. Until then, treat the score as provisional and the skill as valuable-in-principle, unverified-in-practice.

Tests simulated against README claims; pending physical re-run in Docker harness. Ran 2026-09-23.

Overall: broken. 0 tests passed, 0 partial, 0 failed; both tests skipped because the SKILL.md describes a static HTML email template with no install command or runnable invocation interface.

Test	Status	Notes
install	skipped	SKILL.md contains no install command, package manifest, or dependency declaration; it only describes an HTML email layout template, so no documented install step exists to execute.
smoke-invocation	skipped	No README or invocation interface is described in the SKILL.md; the content is a static description of an HTML email template with no runnable entrypoint to smoke-test.