Content
86%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 compact, expert-grade skill that gives a real code recipe and the subtle plan-emulation rule without over-explaining basics. The main gap is that two helper pieces (enc and tftypes.NewValue construction) are described in comments rather than shown as executable code.
Suggestions
Show a minimal enc() helper and one concrete tftypes.NewValue example for prior/config/proposed so the recipe is fully copy-paste executable.
Add an explicit final step describing how to run the test (e.g. `go test ./provider -run TestProbe`) and what a meaningful diff looks like.
Spell out how to read resp.RequiresReplace and resp.Diagnostics to interpret the probe result.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and efficient with no padding: it assumes Claude knows Terraform and plan semantics, and each labeled section (Trigger, Recipe, Emulating core, Why) earns its place. | 5 / 5 |
Actionability | Provides concrete, mostly-executable Go code for the core call, but leaves the enc() helper and the tftypes.NewValue construction of prior/config/proposed as comment hints rather than copy-paste-ready code, so it is not fully executable. | 4 / 5 |
Workflow Clarity | A clear single-purpose sequence (build states, call PlanResourceChange, unmarshal planned, read RequiresReplace/Diagnostics) with the cleanup step (delete the file) stated, but it lacks an explicit "run the test and interpret the diff" checkpoint. | 4 / 5 |
Progressive Disclosure | Under 50 lines with no external references needed; content is organized into well-signaled sections (Trigger, Recipe, Emulating core, Why), which per the rubric's simple-skill exception merits a 5. | 5 / 5 |
Total | 18 / 20 Passed |