Content
75%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, largely actionable agent prompt with a clear four-phase workflow, concrete exemplars, and a defined output format. The main gaps are the absence of a worked output example and an explicit self-verification step, plus a few trimmable rationale sentences.
Suggestions
Add a compact worked example of the output format (one mapped flow, one critical gap with question and default assumption) so the deliverable is unambiguous.
Add an explicit end-of-workflow verification step, e.g., re-checking that every mapped flow was compared against the spec and every gap has a question and default assumption.
Trim rationale sentences that restate the obvious (e.g., 'This prevents generic feedback and surfaces real constraints') and consider whether all three <examples> entries are needed.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and rationale-driven with well-chosen good/bad exemplars, but includes a few trimmable rationale sentences (e.g., "This prevents generic feedback and surfaces real constraints") and a three-example block that adds length. Mostly lean with minor over-explanation, fitting anchor 4; it does not reach anchor 5's every-token-earns-its-place bar. | 4 / 5 |
Actionability | As an instruction-only skill, the guidance is concrete: named tools (Grep, Glob), explicit good-vs-bad question examples, a structured output format with severity tiers, and actionable heuristics ("Gaps are only gaps if the codebase doesn't already handle them"). It falls short of anchor 5 because there is no worked example of an actual analyzed flow or a filled-in output to anchor the format. | 4 / 5 |
Workflow Clarity | Four clearly sequenced phases with an explicit grounding loop (Phase 1 codebase context feeds Phase 3's "don't flag it as a gap" rule). The task is analytical and non-destructive, so the feedback-loop cap does not apply, but there is no explicit verification checkpoint (e.g., confirming every mapped flow was checked), placing it at anchor 4 rather than 5. | 4 / 5 |
Progressive Disclosure | No bundle files exist, and the single ~100-line file is well-sectioned with clear headers and no nested references. Content is appropriately placed in one file, though the leading examples block could arguably live in a separate reference; this fits anchor 4's "good structure, minor organization gaps" rather than the well-signaled-references structure of anchor 5. | 4 / 5 |
Total | 16 / 20 Passed |