Content
62%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill is highly actionable with excellent workflow clarity and concrete templates, but suffers significantly from verbosity. The Review Workload Forecast section alone accounts for a large portion of the content and contains substantial redundancy (table format + plain-text lines + prose explanation + return summary all covering the same fields). The task writing rules and phase organization guidelines, while useful, explain concepts Claude already understands and could be dramatically condensed.
Suggestions
Cut the Review Workload Forecast section by ~50%: remove the prose re-explanation of fields already shown in the template, and consolidate the table + plain-text lines into a single format specification with a note about the guard contract.
Remove the Task Writing Rules table — Claude already knows what 'specific', 'actionable', 'verifiable', and 'small' mean. Replace with a single line: 'Each task: one file or logical unit, with concrete file path and verifiable outcome.'
Move the Phase Organization Guidelines tree to a referenced file (e.g., `_shared/phase-templates.md`) since it's supplementary context, not core workflow.
Condense the chain strategy explanations (stacked-to-main, feature-branch-chain, size-exception) into a compact table rather than a numbered prose list with full sentences.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is extremely verbose at ~300+ lines with significant redundancy. The Review Workload Forecast rules are explained in prose, then repeated in the template, then repeated again in the return summary. The task writing rules table, phase organization guidelines, and multiple format specifications bloat the content well beyond what's necessary. Much of this (like explaining what 'specific' vs 'actionable' means) is knowledge Claude already has. | 1 / 3 |
Actionability | The skill provides highly concrete, executable guidance: exact file paths, specific markdown templates with copy-paste-ready formats, explicit field values, clear decision trees for chain strategies, and detailed examples of good vs bad tasks. Every step has specific outputs and formats. | 3 / 3 |
Workflow Clarity | The 5-step workflow is clearly sequenced (Load Skills → Analyze Design → Write tasks.md → Persist Artifact → Return Summary) with explicit validation checkpoints. The Review Workload Forecast includes a decision tree with clear branching logic based on delivery strategy, and the persistence step is marked as MANDATORY with explicit instructions. | 3 / 3 |
Progressive Disclosure | The skill references shared files like `skills/_shared/sdd-phase-common.md` and `skills/_shared/openspec-convention.md` appropriately, but no bundle files are provided to verify these exist. The main content itself is monolithic — the lengthy Review Workload Forecast rules, phase organization guidelines, and task writing rules could be split into referenced files rather than inlined, making the core skill leaner. | 2 / 3 |
Total | 9 / 12 Passed |