Content
67%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.
The body is well structured and highly actionable, with concrete rule tables, a mechanical scan checklist, a disciplined process with genuine checkpoints, and a real linter script. Its weaknesses are redundancy (the linter and the modes/split are each explained multiple times, and the licensing paragraph runs long) and one broken reference to a nonexistent examples file.
Suggestions
Document ste-lint.py once — fold the flag-by-flag detail from Process step 3 into the Additional Resources entry (or vice versa) so the linter's behavior is stated in one place.
Trim the Source and Scope licensing paragraph to one sentence ('the standard's license forbids redistribution, so the dictionary is not included; request it from the official downloads page'), dropping the quoted clause and the eight exempt categories.
Create examples/before-after.md (the worked examples the body promises) or remove the reference to it — as written, two sections point to a file that is not in the bundle.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is tight sentence-by-sentence but over budget overall: the linter's flags and options are documented twice with overlapping detail (Process step 3 and Additional Resources both cover --baseline, --disable, synonym rotation, and never-flagging hedges), the Source and Scope section spends a full paragraph on ASD licensing details (quoting page 2 and enumerating eight exempt categories), and the modes/structural-lexical split is explained three separate times. This matches anchor 3 ('mostly efficient but includes some unnecessary explanation or could be tightened') better than 4. | 3 / 5 |
Actionability | Concrete Do/Don't rule tables, a Scan Checklist with example phrases and fixes, a precise Output Format with an example table, and a real executable script with flags give mostly executable guidance. Minor gaps: no literal sample invocation of ste-lint.py (its usage is described, not shown), and the worked-examples file the body relies on (examples/before-after.md) does not exist — keeping this at anchor 4 rather than 5. | 4 / 5 |
Workflow Clarity | The Process gives a clear six-step sequence with real checkpoints: read for meaning before rewriting, the modality check before committing to a rewrite, flag-instead-of-silently-simplify error handling, and 'if the input already complies, say so'. This is an instruction-only rewriting skill, so the destructive/batch cap does not apply. The gap is the absence of an explicit post-rewrite verification pass (e.g., re-running the linter or checklist on the output), which holds it at anchor 4 rather than 5. | 4 / 5 |
Progressive Disclosure | The body is a well-organized overview with one-level-deep, clearly signaled references (inline cross-references plus a consolidated Additional Resources section), and the two existing bundle files (references/writing-rules.md, scripts/ste-lint.py) are substantive and correctly placed. However, one of the three referenced paths — examples/before-after.md, cited twice in the body — does not exist in the bundle, a concrete navigation defect that fits anchor 4 ('minor organization gaps') rather than 5. | 4 / 5 |
Total | 15 / 20 Passed |