Content
67%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, actionable workflow orchestrator with clear sequencing and checkpoints, weakened mainly by repeated exhortations and duplicated content between the two modes.
Suggestions
Factor the shared steps of nori-full-send and nori-copilot into a common section to remove duplication and tighten the token budget.
Remove the repeated 'DO NOT BE LAZY' / 'CRITICAL add each element' boilerplate — state the TodoWrite requirement once at the top.
Trim the Tone section's anecdotal padding ('My last assistant was too sycophantic...') to the behavioral directives only.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient directive prose, but the 'CRITICAL ... DO NOT BE LAZY' exhortation is repeated three times, the two mode sections duplicate many steps, and the Tone section includes padding ('My last assistant was too sycophantic and was replaced...'). | 3 / 5 |
Actionability | Steps are concrete and executable — 'Check git status - are you on main, master, dev...', 'Invoke the using-skills skill', 'Ask me to pick a mode' — with only minor gaps like the vague 'Derive the branch name from my request'. | 4 / 5 |
Workflow Clarity | A clearly sequenced checklist workflow with checkpoints (protected-branch gate, plan-approval feedback loop 'Repeat until I approve'), though most validation is delegated to sibling skills rather than stated inline. | 4 / 5 |
Progressive Disclosure | Well-organized into labeled sections (Full-send Mode, Copilot Mode, Tone, Coding Guidelines) with a clean one-level 'See also' pointer list to sibling skills and no nested references; the duplicated mode structure is the only minor organization gap. | 4 / 5 |
Total | 15 / 20 Passed |