Content
88%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.
The body is a well-structured, highly actionable orchestration protocol: explicit MUST rules, concrete coordinator dialogue, sequenced workflows with verification checkpoints and deadlock escalation, worked examples, and an anti-pattern checklist. Its only real weaknesses are the dense bolted-on Waza/Rai boundary paragraphs, which slightly hurt conciseness and organization.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dominated by terse imperatives ("The original author is locked out. They may NOT produce the next version of that artifact. No exceptions.") with no explanation of concepts Claude already knows, matching the "efficient; minor instances of over-explanation" anchor. It misses anchor 5 because the dense "Waza rejection boundary (Squad only)" and "Rai/Fact Checker output boundary" paragraphs read as appended policy blocks with some redundancy, and the Context section restates the Patterns premise ("This prevents defensive feedback loops..."). | 4 / 5 |
Actionability | The guidance is fully executable as behavior: explicit MUST directives ("the Coordinator MUST verify that the selected agent is NOT the original author"; "the Coordinator MUST refuse and ask the Reviewer to name a different agent"), ready-to-use coordinator dialogue ("The original author is locked out. Please name a different agent."), and four worked examples covering the common cases (reassign, escalate, deadlock, reviewer mis-assignment). Per the scoring notes, the absence of code in an instruction-only skill is not penalized when the guidance is this actionable. | 5 / 5 |
Workflow Clarity | The rejection → lockout → revision → rereview → approval sequence is explicit throughout, with a pre-spawn validation checkpoint ("Before spawning a revision agent, the Coordinator MUST verify that the selected agent is NOT the original author"), a focused rereview gate before approval, and an error-recovery escalation path for deadlock ("the Coordinator MUST escalate to the user rather than re-admitting a locked-out author"). The examples walk each full sequence step by step. | 5 / 5 |
Progressive Disclosure | Sections (Context, Patterns with subsections, Examples, Anti-Patterns) are clearly headed and easy to navigate with no nested references, matching the "good structure; minor organization gaps" anchor. It does not reach anchor 5 because the skill exceeds 50 lines yet inlines two dense, tangential sub-policy paragraphs ("Waza rejection boundary (Squad only)", "Rai/Fact Checker output boundary") mid-section that would be better placed in their own signaled sections or a separate reference file. | 4 / 5 |
Total | 18 / 20 Passed |