Content
81%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-engineered, instruction-only skill body: it prescribes an unambiguous four-phase workflow with explicit validation and failure-reporting checkpoints, and its guidance is concrete down to exact tool calls, filter syntax, and config keys. The only real costs are mild repetition of the setup-incomplete/repository-filter rules across routes and the absence of the search tool's exact call signature.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and procedural with no explanation of concepts Claude already knows; every section gives operational rules. It is not a 5 because the coding-session repository-filter rule and the "report setup is incomplete" instruction are each repeated nearly verbatim across the three routes and could be consolidated into the Resolve Configuration section. | 4 / 5 |
Actionability | Gives concrete, executable guidance: exact calls like `bm_project(command="list")`, `bm_read` with `project=`, literal filter syntax `filters={"repository": "<configured repository>"}`, named config keys, and concrete note types and statuses. It falls short of anchor 5 because the search tool is only referred to as "note/search calls" without its exact name or parameter signature, leaving one common case not fully copy-paste ready. | 4 / 5 |
Workflow Clarity | A clear four-phase sequence (Resolve Configuration → Choose One Recall Route → Check Current State → Present and Continue) with explicit decision points: the 0/1/multiple match handling ("show at most three ... then wait for the user to choose"), the drift-comparison matrix keyed on repository/SHA/branch/cwd, and explicit instructions to report unprovable comparisons. Validation checkpoints and error-recovery reporting are present throughout, matching the anchor-5 example's structure. | 5 / 5 |
Progressive Disclosure | The body is self-contained with well-organized, clearly headed sections and no nested or buried references; the only bundle asset (assets/icon.svg) is appropriately non-procedural. It is not a 5 because the skill exceeds the ~50-line threshold where a fully inline single file earns the top score, and the per-route rules (exact checkpoint, topic discovery, current repository) are long enough that they could plausibly live in one-level-deep reference files. | 4 / 5 |
Total | 17 / 20 Passed |