Content
57%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 has a distinctive voice and a concrete, well-structured command surface, but token budget is spent on persona padding, workflows are described as fragments rather than sequenced steps, and several referenced files are missing from the bundle. Tightening the marketing prose and fixing the dangling references would lift it substantially.
Suggestions
Delete the Keywords section — trigger matching is the description's job; a keyword list in the body burns tokens without adding executable guidance.
Give each command an explicit ordered workflow with checkpoints instead of fragment descriptions (e.g., /em:postmortem: 1. timeline of failure, 2. 5 Whys, 3. separate contributing factors from root cause, 4. assign owner + verification date per change).
Resolve the dangling references: either include agents/devils-advocate.md and the five ../<skill>/SKILL.md targets in the bundle, or remove/inline that guidance so every referenced path resolves.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The core sections (Commands, Commands in Detail, Reasoning Technique) are lean, but several sections are padding that adds no executable guidance: the Keywords list (trigger terms belong in the description), 'The Difference', 'What This Isn't', 'This isn't therapy. It's preparation.', and the Andy Grove/Ben Horowitz anecdote. | 3 / 5 |
Actionability | Guidance is mostly executable: a command table with five concrete slash commands, a Quick Start with runnable 'python scripts/decision_matrix_scorer.py' invocations, and named frameworks per command (reversibility test, 10/10/10, 5 Whys). Minor gaps: no worked example of what command output (e.g., a severity-rated vulnerability list) should look like. | 4 / 5 |
Workflow Clarity | Each command is described in 2-4 sentence fragments ('5 Whys without softening, contributing factors vs root cause, owners per change, verification dates') rather than an explicit sequenced workflow, and validation is limited to the single Internal Quality Loop in the Communication section. The sequence is present but checkpoints are implicit. | 3 / 5 |
Progressive Disclosure | The three references/ and two scripts/ files are real, well-signaled, and one level deep, but the body also references 'agents/devils-advocate.md' which does not exist in the bundle, plus five unverifiable cross-skill paths ('../challenge/SKILL.md', '../board-prep/SKILL.md', etc.) and 'c-level-advisor/skills/agent-protocol/SKILL.md' — over half the referenced paths dangle. | 3 / 5 |
Total | 13 / 20 Passed |