Content
81%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 tight, code-first skill that is immediately usable: executable functions, an email-type template map, and a quick-reference table with no token waste on concepts Claude already knows. The main flaws are a latent f-string bug in the system prompt, a hardcoded model version, and no guidance for handling truncation on longer drafts.
Suggestions
Fix the system-prompt f-string so the length mapping resolves to a real value (e.g. compute length_desc = {'short': '2-3 sentences', ...}.get(length, '2-3 paragraphs') before building the prompt) instead of injecting literal dict-lookup syntax into the model's instructions.
Move the model ID out of the hardcoded 'claude-opus-4-6' into a constant or parameter so it does not go stale as models are deprecated, and bump max_tokens (or scale it with length) so 'long' full-email drafts are not truncated at 1024 tokens.
Trim the duplicated length documentation between the function-signature comment and the system prompt so the parameter is specified in exactly one place.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes Claude's competence — no padding or re-teaching of email concepts — but the system-prompt f-string re-specifies the length parameter already documented in the function signature, and a hardcoded model version ("claude-opus-4-6") adds time-sensitive detail. These are minor trims, matching anchor 4 rather than the every-token-earns-its-place of 5. | 4 / 5 |
Actionability | Provides mostly executable, copy-paste-ready functions with defaults and prompt templates covering draft, type-specific draft, and improve flows. However, the f-string at the system-prompt line emits literal `{"short": ...}.get(length, ...)` text into the prompt instead of the resolved value, and max_tokens=1024 can truncate 'long' full emails — concrete code with minor gaps, not the fully robust coverage of a 5. | 4 / 5 |
Workflow Clarity | This is a simple, single-purpose skill (three unambiguous API-call functions, no multi-step or destructive/batch operations), so the simple-skill exception applies: each action is unambiguous and the section structure makes the right entry point obvious. There is nothing risky that would demand validation checkpoints, so it does not fall to the anchor-3/4 territory. | 5 / 5 |
Progressive Disclosure | The single file is well-organized with clear section headers, an overview, and a quick-reference table, with no nested or buried references. At roughly 99 lines it exceeds the under-50-line bar that would qualify a compact single file for a 5, so it sits at 'good structure; most content appropriately placed' — a 4. | 4 / 5 |
Total | 17 / 20 Passed |