Content
67%Scale 1-5Reviews 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 meta-skill that effectively serves as a routing layer for discovering and applying specific engineering workflow skills. Its main strengths are the clear decision tree for skill discovery and the lifecycle sequence showing how skills compose. Its main weakness is verbosity — the Core Operating Behaviors and Failure Modes sections explain general engineering principles that Claude already knows, and the quick reference table largely duplicates the discovery tree, inflating the token cost of this frequently-loaded skill.
Suggestions
Trim or remove the 'Core Operating Behaviors' section — these are general engineering principles Claude already knows, and they significantly inflate the token cost of a skill that's loaded at session start
Remove the 'Failure Modes to Avoid' section as it largely restates the Core Operating Behaviors in negative form
Eliminate either the discovery tree or the quick reference table, as they convey largely the same information; keep the tree since it's more actionable for routing decisions
Add hyperlinks to referenced skills (e.g., `[spec-driven-development](skills/spec-driven-development/SKILL.md)`) so Claude can navigate directly
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill contains useful content but is notably verbose in places. The 'Core Operating Behaviors' section extensively explains concepts Claude already understands (don't be sycophantic, surface assumptions, push back). The 'Failure Modes to Avoid' section largely repeats the Core Operating Behaviors. The quick reference table duplicates the discovery tree. However, the discovery tree itself and the lifecycle sequence are efficient and valuable. | 3 / 5 |
Actionability | The skill discovery tree is highly actionable — given a task type, Claude knows exactly which skill to invoke. The assumption-surfacing template is concrete and copy-paste ready. However, the skill is primarily a routing/meta-skill, so 'actionability' manifests as clear decision-making rather than executable code, which it does reasonably well. Minor gap: no concrete example of how to actually invoke/load a skill file. | 4 / 5 |
Workflow Clarity | The lifecycle sequence provides a clear ordering of skills for a complete feature, and the note that not every task needs every skill is helpful. The discovery tree provides clear routing logic. However, there's no explicit validation checkpoint in the meta-workflow itself — e.g., 'after selecting a skill, confirm it matches before proceeding' — though the referenced skills presumably contain their own validation steps. The Definition of Done reference adds a verification layer. | 4 / 5 |
Progressive Disclosure | The skill references many sub-skills by name and references `references/definition-of-done.md`, creating a clear one-level-deep structure. The discovery tree and quick reference table serve as navigation aids. However, no bundle files are provided to verify the referenced skills exist, and the references aren't hyperlinked (e.g., `[spec-driven-development](skills/spec-driven-development/SKILL.md)`). The content is also somewhat long for an overview/routing skill — the Core Operating Behaviors section could arguably live in a separate referenced file. | 4 / 5 |
Total | 15 / 20 Passed |