Content
85%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.
The body is a strong, dense operational guide: fully sequenced workflow with approval and retry safeguards, lean prose, and a clean split between overview and bundled scripts. The only weaknesses are minor — no worked input_json example and a slightly padded opening sentence.
Suggestions
Add one worked `input_json` example for a common case (e.g. background swap with image_urls plus a change-X-keep-Y prompt) so the quote/create commands can be adapted without a schema round-trip.
Trim the opening definitional sentence about image-to-image editing — Claude already knows this — and lead directly with the PopNow contract and workflow.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and almost entirely operational, but has minor trimmable instances: the opening definitional sentence ("Image-to-image editing takes one or more input images plus an instruction and returns a new image") explains a concept Claude already knows, and the versioned model paths (e.g. nano-banana-2.1, seedream-5.0-pro) are time-sensitive — mitigated by the "confirm them with `models`" instruction but not isolated in a deprecated/old-patterns section. This fits the efficient-with-minor-trims anchor rather than the every-token-earns-its-place anchor. | 4 / 5 |
Actionability | All four workflow steps have concrete executable commands ("python3 scripts/popnow.py models --search "edit image"", quote/create/status invocations) with precise argument semantics (quoteId, approved decimal string up to nine places, idempotency-key reuse). It stops short of the top anchor because no worked `input_json` example is given — the reader must derive the object shape entirely from the schema lookup, a minor gap for a common case like a background swap. | 4 / 5 |
Workflow Clarity | The four-step sequence has explicit validation checkpoints and error-recovery feedback loops: schema confirmation before building input, mandatory approval of the exact max USD ("a zero-price quote still needs approval"), idempotent retry rules ("retry only with the same key and payload", "Do not retry terminal failures"), pending-is-not-failure guidance, and handling for expired quotes, auth failure and low balance. This matches the anchor requiring explicit validation steps plus feedback loops. | 5 / 5 |
Progressive Disclosure | The SKILL.md is a well-organized overview and executable detail is appropriately split into the bundled helper (scripts/popnow.py, verified to exist and to import scripts/client.py), which the body references directly; nothing that belongs in a separate file is inlined, and the compact model table with best-for guidance is overview-appropriate. Navigation is easy with clear section headers and one-level-deep structure. | 5 / 5 |
Total | 18 / 20 Passed |