Content
78%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 lean, well-ordered behavioral instruction set that assumes Claude's competence and gives unusually specific pacing and sequencing directives for speech-plus-board-work interaction. Its main weakness is density: several bullets fuse multiple rules into single overloaded sentences, and a couple of abstract rules would land better with a one-line good/bad example.
Suggestions
Split the overloaded multi-rule bullets (arrow semantics, claim consistency) into shorter one-rule lines so each instruction is independently checkable.
Add a one-line good/bad example for the abstract rules, e.g. a label that weakens a claim vs. one that preserves its conditions.
Make the pre-close verification explicit (a final check that every drawn element was explained and no omitted element is referenced) to complete the workflow's feedback loop.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is ~13 dense bullets with zero padding and no explanations of concepts Claude already knows — every bullet encodes non-obvious policy such as "Do not silently plan the whole diagram before speaking". It falls short of a 5 because several bullets (the arrow-semantics and claim-consistency rules) each pack 4-5 distinct rules into one overloaded sentence that could be split or trimmed. | 4 / 5 |
Actionability | Most rules are concrete, executable directives: "finish one meaningful board block, then output that block's concise spoken explanation before starting the next block's drawing calls", "identify the nodes and connection points before placing them", "dispatch it immediately before the sentence that discusses that element". As an instruction-only skill this is appropriately code-free, but a few bullets remain abstract ("Preserve the conditions, reference frame, units, and comparison basis needed for each claim") with no example of a compliant vs. non-compliant board block, keeping it below 5. | 4 / 5 |
Workflow Clarity | A clear sequence is present: answer aloud first, decide on board use, loop draw-block-then-explain per block, time spotlights to sentences, finish the board before requesting close. Checkpoints exist ("Check the planned text and geometry against the sentence they illustrate. Correct a conflicting element before drawing it"; premise-checking before repeating a Director's task), matching 'clear sequence with most checkpoints present'. Not a 5 because a couple of checkpoints are implicit — e.g., no explicit final verification pass before close beyond the one-line reminder. | 4 / 5 |
Progressive Disclosure | The skill is under 50 lines, single-purpose, with no need for external references; the flat, logically ordered bullet list is well-organized with each bullet covering one concern. Per the rubric's simple-skill guidance, this earns a 5 — nothing here belongs in a separate reference file, and there are no nested or buried references. | 5 / 5 |
Total | 17 / 20 Passed |