Content
85%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a well-structured, highly actionable skill that handles significant complexity (10+ providers, multiple generation modes, extensive configuration) with clear workflows and excellent progressive disclosure. Its main weakness is moderate verbosity — the identity-preservation guidance appears twice, the Codex OAuth section is lengthy, and some explanatory passages could be tightened to save tokens. Overall it's a strong skill that effectively balances comprehensiveness with navigability.
Suggestions
Consolidate the two identity-preservation sections (under 'Usage' and the standalone 'Reference-Image Identity Preservation' section) into a single section to eliminate redundancy.
Condense the Codex OAuth explanation — the core message ('Codex login ≠ OPENAI_API_KEY; use --provider codex-cli instead') can be stated in 3-4 lines with a reference link for details, rather than the current ~15 lines.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is quite long and includes some redundant sections (e.g., identity-preservation guidance is stated twice in slightly different forms, the Codex OAuth explanation is verbose and could be condensed). However, much of the content is genuinely necessary given the complexity of supporting 10+ providers. Some sections like the environment variables table are exhaustive but arguably needed as reference. | 2 / 3 |
Actionability | The skill provides fully executable CLI commands with concrete examples for every major use case (basic generation, reference images, batch mode, provider-specific invocations). The options table, environment variables, and model resolution priority are all specific and copy-paste ready. | 3 / 3 |
Workflow Clarity | The workflow is clearly sequenced: Step 0 (blocking config load) → provider selection → model resolution → generation. The batch mode has explicit validation (retry up to 3 attempts, success/failure counts). Error handling covers missing keys, invalid ratios, and unsupported provider/model combos with specific fallback paths. The generation mode decision table clearly guides when to use sequential vs batch vs subagents. | 3 / 3 |
Progressive Disclosure | The skill has excellent progressive disclosure with a clear overview in the main file and well-signaled one-level-deep references to provider-specific guides, config schema, setup flow, usage examples, and fallback documentation. The references table at the bottom provides a clean navigation index. Content is appropriately split between the main skill and reference files. | 3 / 3 |
Total | 11 / 12 Passed |