Content
73%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.
Highly actionable and workflow-sound content: exact commands, config resolution rules, validation, and iteration loops are all present. Its main flaw is token inefficiency — the same reporting/save-path rules are restated in up to three places, and prompt-taxonomy detail is duplicated inline despite dedicated reference files.
Suggestions
Consolidate the repeated save-path/reporting rules ('attach the generated PNG from generated-images/', 'do not print the final prompt/model/base URL', 'do not attach AGENTS.md') into a single stated-once section; they currently appear across Save-path policy, Workflow steps 11/14, and the Commands aftercare block.
Move the use-case taxonomy slugs, shared prompt schema, and worked examples into references/prompting.md or references/sample-prompts.md and keep only a short summary plus pointer in SKILL.md, matching the existing Reference map pattern.
Merge the intro paragraph and the 'Top-level modes and rules' block, which both explain that this skill mirrors the official imagegen skill except for the execution layer.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body systematically repeats guidance across sections: 'report and attach the generated PNG file from generated-images/' appears three times (Save-path policy, Workflow step 11, Commands aftercare), 'do not print the final prompt/model/base URL' twice (Workflow step 14 and Commands), and 'do not attach unrelated files such as AGENTS.md' twice. It also re-explains its relationship to the official skill both in the intro and the rules block. | 2 / 5 |
Actionability | Fully executable: exact PowerShell CLI invocations with flags (--size, --out-dir, --quality, --background), exact env vars and config reads (CODEXMANAGER_IMAGE_BASE_URL, config.toml provider base_url, auth.json key), exact endpoint paths and request fields, and concrete post-call parsing guidance ('Use the first absolute paths[] item as the final image path'). | 5 / 5 |
Workflow Clarity | The 14-step workflow is clearly sequenced with an explicit validation checkpoint ('Inspect outputs and validate: subject, style, composition, text accuracy, and invariants/avoid items') and a feedback loop ('Iterate with a single targeted change, then re-check'), plus explicit batch strategy and non-destructive versioned-save rules for edits. | 5 / 5 |
Progressive Disclosure | A dedicated 'Reference map' clearly signals one-level-deep references, and all six referenced bundle files exist. However, roughly a hundred lines of use-case taxonomy, shared prompt schema, and example specs are inlined even though references/prompting.md and references/sample-prompts.md exist to hold that detail, keeping it below the top anchor. | 4 / 5 |
Total | 16 / 20 Passed |