Content
73%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 efficiently documents a hook-delivered, model-no-op skill with a clear single action and one well-signaled external reference. Actionability is inherently limited because the model takes no action, and the absence of section structure keeps progressive disclosure just below the top anchor.
Suggestions
Add brief section headers (e.g. "## What this skill does" and "## Output fields") to lift progressive_disclosure from 4 to 5.
Surface the rule-overhead/net output behavior in the description so specificity reflects the skill's full feature set.
Include a few natural user-side trigger phrases (e.g. "token cost", "tokens used", "session spend") alongside the slash command to broaden trigger-term coverage.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly lean and assumes competence (hook filenames, env var, default token value stated directly), with only minor editorial padding such as "rather than hiding the net-negative regime behind a gross-savings number", fitting the efficient level-4 anchor over the more padded level-3. | 4 / 5 |
Actionability | Concrete specifics exist (the CAVEMAN_RULE_OVERHEAD_TOKENS override, the 1,250 tokens/turn default, named output lines, referenced hook files), but because "The model does not need to do anything when this skill fires," the guidance is largely explanatory rather than executable, sitting at level 3 rather than the mostly-executable level 4. | 3 / 5 |
Workflow Clarity | This is a simple single-action skill whose one action is stated unambiguously ("the hook returns decision: block with the formatted stats as the reason. The user sees the numbers immediately"), qualifying for the simple-skill exception; it is not a destructive/batch operation so the level-3 validation cap does not apply. | 5 / 5 |
Progressive Disclosure | The short body is a concise overview with one clearly signaled one-level reference ("(see docs/HONEST-NUMBERS.md)"), but it lacks section headers (just two prose paragraphs), so it lands at the well-structured level 4 rather than the cleanly sectioned level 5. | 4 / 5 |
Total | 16 / 20 Passed |