Content
85%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is highly actionable with executable tmux/PTY harnesses, profiling recipes, and a clear sequenced loop with checkpoints. The only real weakness is mild redundancy between the description's triggers and the "What It Is Used For" section.
Suggestions
Trim overlap with the description: compress the "What It Is Used For" list to cases not already named in the frontmatter "Use for" clause, or fold it into the intro sentence.
Add one explicit error-recovery note in the Harness Loop (e.g., what to do if the expected screen pattern never appears) to strengthen the existing checkpoint.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient with no concept explanations Claude already knows, but the "What It Is Used For" list and intro restates several triggers from the description (startup regressions, memory leaks, hangs, prompt flows, terminal demos) and could be tightened. It is leaner than the verbose level-1 anchor but does not fully meet "every token earns its place." | 2 / 3 |
Actionability | Executable tmux and PTY code blocks plus concrete profiling recipes give copy-paste-ready guidance; placeholders like <command-under-test> are explicitly justified by the guardrails (adapt to the current repo), so this is not mere pseudocode. It clears the level-2 pseudocode bar. | 3 / 3 |
Workflow Clarity | The 8-step Harness Loop is clearly sequenced with an explicit checkpoint ("Wait for a concrete screen pattern or prompt before the next action") and a clean-kill final step. Because driving a CLI is not a destructive/batch operation, the absence of a validate-retry loop does not cap the score. | 3 / 3 |
Progressive Disclosure | It is a single well-organized file with clear sections and no nested/deep references; no bundle files exist, and the scope fits one file. This matches the level-3 well-organized-sections guidance for skills that need no external references. | 3 / 3 |
Total | 11 / 12 Passed |