Content
77%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.
An exceptionally actionable, well-sequenced wizard whose commands, decision tables, and error-recovery paths are copy-paste ready, with validation and resume logic that fully de-risks the mutating Dataverse operations. Its weaknesses are token weight (boilerplate repetition and historical rationale prose) and a monolithic structure that inlines detail the referenced files should carry.
Suggestions
Move the Step 9 canonical offline-profile.json schema, schema invariants, and field-rationale table into offline-profile-schema.md, keeping only the field map and the illustrative example inline.
Extract the publish error recipes (0x80071141 cycle handling, 0x80071140 fallback, PublishAllXml timeout protocol) into dataverse-offline-api.md and keep one-line pointers plus the decision rules in the body.
Trim the repeated `cd -- '<working_dir>' || { echo "BLOCKED…"; exit 1; }` guard (state the binding once in the App-root binding section) and condense design-rationale/history blockquotes and empirical date stamps into a short changelog or old-patterns section.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense operational instruction with no teaching of concepts Claude already knows, but it is padded by a repeated `cd -- '<working_dir>' || …` guard in every code block, multi-paragraph design-rationale/history blockquotes (e.g. the Step 3.5 rationale and the "chanel-rm and FCB Tracker" note), and inline empirical date stamps ("Empirical 2026-05-24") — more than minor over-explanation, so it sits at the midpoint rather than 4. | 3 / 5 |
Actionability | Every step ships copy-paste-ready bash via the dataverse-request.js wrapper with full JSON bodies, exact OData queries, decision tables mapping conditions to actions, and specific error codes (0x80071141, 0x80071140) with concrete recovery commands — fully executable and covering common failure cases. | 5 / 5 |
Workflow Clarity | The 10-step workflow is explicitly sequenced with per-step telemetry checkpoints, an interactive approval flow, a dedicated verify step (9.5) with drift handling, pre-publish cycle detection, resume/idempotency logic, and a failure-recovery table — explicit validation and feedback loops throughout a mutating batch operation. | 5 / 5 |
Progressive Disclosure | References are clearly signaled at the top, but the skill is a ~800-line monolith that inlines material belonging in the referenced files (the full canonical offline-profile.json schema plus a field-rationale table despite a schema reference existing, publish error recipes, timeout protocols), and no local bundle files (references/, scripts/, assets/) exist — above the "buried references" anchor but with substantial inline content that should be split out. | 3 / 5 |
Total | 16 / 20 Passed |