Content
57%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 highly actionable, code-heavy reference for Whisper with accurate API usage across Python and CLI. Its main weaknesses are an unreferenced bundle file whose content is partially duplicated inline, a batch-processing example lacking validation, and a few padded sections (GitHub-star metrics, performance tables).
Suggestions
Replace the inline 'Language support' section with a one-line pointer to references/languages.md (e.g., 'See [languages.md](references/languages.md) for the full 99-language list, WER by model size, and language-specific tips') so the bundle file is actually discoverable and the duplication is removed.
Add validation to the batch processing example: wrap the transcribe call in try/except, check that each input file exists before processing, and confirm the output file was written before moving on.
Trim the 'Metrics' block (GitHub star counts are time-sensitive and not actionable) and drop or shrink the CPU/GPU real-time-factor performance table, which largely repeats the speed column already shown in the model sizes table.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient, code-first content, but the inline 'Language support' section ('Top-supported languages... Full list: 99 languages total') duplicates references/languages.md, and the 'Metrics' block ('72,900+ GitHub stars') and the CPU/GPU real-time-factor performance table add little actionable value. This is more than anchor 4's 'minor instances' of padding but far from anchor 2's heavily padded style. | 3 / 5 |
Actionability | Nearly all guidance is copy-paste ready: 'pip install -U openai-whisper', 'whisper audio.mp3 --output_format srt', and model.transcribe() examples with language, task, initial_prompt, word_timestamps, and temperature options. It falls short of anchor 5 because the LangChain section ('from langchain.document_loaders import WhisperTranscriptionLoader') references a loader that does not appear to exist and uses a deprecated import path, so that example is not reliably executable. | 4 / 5 |
Workflow Clarity | The core single action (load model, transcribe) is unambiguous, but the batch processing section loops over files writing outputs with no validation or error handling, and the rubric explicitly caps workflow clarity at 3 for batch operations without verification steps. It is not anchor 2 because the sequences that are present are coherent and well-ordered. | 3 / 5 |
Progressive Disclosure | The body is well-sectioned with headers, but the one bundle file (references/languages.md) is never referenced or linked anywhere in SKILL.md — the 'Language support' section says 'Full list: 99 languages total' with no pointer to the file — and its content is partially inlined instead. This matches anchor 3 ('references present but not clearly signaled; content that should be separate is inline') rather than anchor 4's 'references mostly clear'. | 3 / 5 |
Total | 13 / 20 Passed |