Content
78%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 minimal, well-structured probe template that is highly token-efficient and clearly sequenced with a verify step. Its main weakness is that the actionable fields are placeholders rather than concrete worked examples.
Suggestions
Add one short worked example showing the template filled in for a representative traceback (e.g. a KeyError) so the placeholder fields have a concrete model to follow.
Include an explicit validate-fix-retry loop note (e.g. "run VERIFY; if it fails, revise PATCH_TARGETS and re-run") to turn the single VERIFY step into a feedback loop.
Tighten the CHECKS and PATCH_TARGETS placeholders with 1-2 concrete instance hints per field (e.g. example guard-clause or pydantic validation patterns) to lift actionability.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Extremely lean — a bare template plus two command-shape hints with zero padding or explanation of concepts Claude already knows; every token earns its place. | 5 / 5 |
Actionability | Provides a clear output template and concrete command shapes ("pytest -k <symbol>", "python -m pytest <path>"), but the CHECKS/REPRODUCER/PATCH_TARGETS fields are placeholder-driven rather than executable examples, leaving key specifics to be filled in. | 3 / 5 |
Workflow Clarity | The template imposes a clear CHECKS → REPRODUCER → PATCH_TARGETS → VERIFY sequence with an explicit VERIFY checkpoint and a safety note ("Do not recommend destructive commands"), though there is no validate-fix-retry feedback loop. | 4 / 5 |
Progressive Disclosure | Under 50 lines with no need for external references; the template sections are well-organized and self-contained, satisfying the simple-skill exception for progressive disclosure. | 5 / 5 |
Total | 17 / 20 Passed |