Content
85%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.
A well-structured method skill: the six moves are clearly sequenced with validation checkpoints, explicit feedback loops on failure, and a concrete output template, while worked examples are appropriately deferred to real, well-labeled reference files. The main costs are mild redundancy between the moves and the anti-rush rules and some abstract directives that would benefit from a concrete illustration inline.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is efficient and assumes competence — no basic-concept padding — but the anti-rush rules restate constraints already established in the moves ("At least two hypotheses before testing any" repeats move 3; "Not solved until re-observed gone" repeats move 6), so minor trimming is possible, fitting the 4 anchor rather than "every token earns its place". | 4 / 5 |
Actionability | Guidance is concrete and executable for an instruction-only skill: numeric thresholds ("at least two" hypotheses), a falsification-oriented evidence rule, a default stop point ("Propose — do not implement — unless the user asks"), and a copy-paste-ready output template. It stays at 4 rather than 5 because some inherently abstract directives ("escalate to direct observation with more instrumentation") lack concrete examples of what to do, and no worked example is inline. | 4 / 5 |
Workflow Clarity | The six moves are clearly sequenced with explicit validation checkpoints (move 2 confirms the problem is real before cause-hunting; move 6 re-verifies the symptom under the original conditions) and genuine feedback loops for error recovery ("If the symptom persists... return to move 2 with more instrumentation"), plus a checklist-style anti-rush rules section — matching the 5 anchor's "explicit validation steps; feedback loops; checklists". | 5 / 5 |
Progressive Disclosure | The body keeps the core method while offloading heavy end-to-end traces to references, which are one level deep, clearly signaled ("Read one when you are stuck and want a model to follow"), described by content, and both link targets exist as real files in references/. This matches the 5 anchor's clear overview with well-signaled one-level-deep references. | 5 / 5 |
Total | 18 / 20 Passed |