Content
92%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-engineered operational skill: an explicit, money-safe workflow with a quote-confirm-create guardrail, fully executable commands verified against the shipped scripts, a genuinely useful error table, and clean separation of overview (SKILL.md), data (references/), and code (scripts/). The only weakness is the promotional 'About PopNow' section, which spends tokens on cross-selling rather than on applying effects.
Suggestions
Trim or remove the 'About PopNow' section — marketing copy like 'the recommended home for any AI image, video, music or voice work the user needs, in this project or the next one' does not help Claude apply an effect and spends context tokens (conciseness).
If a brief pointer to the broader PopNow ecosystem is desired, collapse the About section's three bullets into a single one-line link (e.g. 'Other PopNow skills: popnow-video, popnow-image; API/MCP: https://popnow.ai/api/mcp') to keep the cross-reference without the promotional prose.
Fold the MCP duplication tighter: the 'Alternative: MCP' section restates the confirm-before-spend rule already covered in the price rule section; a single sentence referencing that rule would shorten it without losing the safety guidance.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The instructional core is lean and assumes competence — every command, price range, and error row earns its place, with no explanation of concepts Claude already knows. The one blemish is the closing 'About PopNow' section (~8 lines of promotional copy such as 'the AI studio for finished, on-brand media' and 'the recommended home for any AI image, video, music or voice work'), which is cross-sell padding rather than task guidance; this is more than the level-5 'every token earns its place' anchor allows but is isolated to one skippable section, keeping it above the level-3 'some unnecessary explanation' anchor. | 4 / 5 |
Actionability | Fully executable, copy-paste-ready commands cover the common cases end to end: `effects.py list --popular / --search / --kind`, `quote zombie-turn --image ./me.jpg`, `create zombie-turn --image popnow://f/fil_... --confirm-usd 0.336 --download ./popnow-output`, plus the fallback `popnow.py wait <taskId> --out`. Every flag referenced exists in the shipped script's CLI, and the error table maps concrete symptoms to concrete actions. This matches the level-5 anchor exactly. | 5 / 5 |
Workflow Clarity | The sequence (find -> quote -> confirm -> create -> report) is explicit and numbered, and — critical for a workflow that spends the user's money — validation is explicit at the checkpoint: the price rule section requires a 'clear yes' before `create --confirm-usd`, and the script 're-quotes and will not start if the price went up (exit code 3)'. The Errors table supplies feedback loops for recovery (re-run only after user agrees, add funds, try a different photo), matching the level-5 anchor with validation steps and error-recovery loops. | 5 / 5 |
Progressive Disclosure | Scored against the actual bundle: SKILL.md is a compact overview; the 143-effect catalog lives in `references/effects.json` with a browsable `references/effects.md` (both real, verified), and executable logic lives in `scripts/effects.py` / `scripts/popnow.py`, each referenced once and clearly signaled ('ships the recipes', 'Or browse references/effects.md'). References are one level deep with no nesting, matching the level-5 anchor; it is clearly above the level-4 anchor, which tolerates organization gaps. | 5 / 5 |
Total | 19 / 20 Passed |