TEST-RESULTS: content-matrix
The full quick-check results of the content-matrix Skill: how it was run, what passed and what failed.
Date: 2026-10-14 · Kind: smoke test, not proof. One run (n=1), one input. No defect rate is claimed.
How: the agent that wrote the skill followed its SKILL.md step by step in the same session, on a fictional small business (a home bakery, «مخبز سنبلة»). It was not run in a fresh Claude session, so automatic triggering was not tested, and the author checking their own skill is a weaker test than an outside run.
| Run | Input | Result |
|---|---|---|
| 1 | examples/input.md: bakery, 3 pillars, 3 posts/week, wants orders, has photos, 2 customer messages, 34 orders in a notebook, a free PDF | examples/expected-output.md: 3 × 6 grid (18 cells), each tagged; 4-week plan of 12 posts matching her stated capacity (8 Carousel, 4 Reels); goal mix 6 / 3 / 3 with the reason for leaving the default; 4 preparation items; 3 assumptions |
What worked: no questions re-asked; the only numbers used were hers (34 orders, ~400 followers); cells without proof were marked «يحتاج دليل» instead of being filled; the comment keyword was tied to a file she said exists; no clock times were invented.
What failed or was fixed: the first run proposed a «healthy bread» angle and a «licensed?» post. It flagged both as needing proof, but the skill had no explicit rule for health/legal claims. Fix after the run: a rule was added (health, legal, safety and financial claims need an official source or her own document; never mention a licence she did not confirm).
Not tested: a fresh Claude session or Claude.ai; the interactive path (asking the 5 questions and waiting); a vague input; 4 pillars; other models.