Content
60%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 core workflow is genuinely executable — real commands, a precise data-format table, and verified bundle scripts — but it is buried in generic boilerplate with duplicated sections, contradictory dependency claims, and documented validation commands ("--help") that the target script does not support. Trimming the template sections and correcting the verification path would move this to a strong skill.
Suggestions
Remove or collapse the generic boilerplate sections (When to Use, Key Features, Output Contract, Deterministic Output Rules, Completion Checklist) and the duplicate Validation Shortcut/Quick Validation pair into one accurate verification step — the body could shrink by half without losing actionable content.
Fix the documented verification commands: forest_survival.py has no --help handling (raw sys.argv parsing), so either document the actual usage error/message or point verification at scripts/validate_skill.py, and replace the fabricated Quick Validation expected output with the script's real output (Survival_forest_{outcome}.png and .csv).
Reconcile factual inconsistencies: Python "3.10+" vs "3.7+ recommended", "plus 1 additional script(s)" (there are 2), and the "See `## Workflow` above" cross-reference that points to a section defined below.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Roughly 220 lines of which a large fraction is generic template boilerplate ("When to Use", "Key Features", "Output Contract", "Deterministic Output Rules", "Completion Checklist") that adds no skill-specific information, plus two near-duplicate validation sections and contradictory Python requirements ("3.10+" in Dependencies vs "3.7+ recommended" later). This matches anchor 2 (noticeably verbose, several padded sections) rather than 1, because genuinely useful skill-specific content (data-format table, commands, output spec) is also present. | 2 / 5 |
Actionability | The workflow section gives copy-paste-ready commands with full argument syntax for both R and Python ("Rscript scripts/forest_survival.R \"<csv_path>\" \"<outcome_name>\" \"<output_dir>\""), a concrete data-format table with example values, and dependency install commands — all verified to exist in the bundle. It falls short of 5 because "python scripts/forest_survival.py --help" is documented twice but the script parses raw sys.argv and has no --help handling (it would error treating --help as the CSV path), and the Quick Validation "expected output" ("Result file: meta_forest_model_plot_result.md...") is fabricated — the script produces a usage line and Survival_forest_*.{png,csv} files. | 4 / 5 |
Workflow Clarity | The three-step workflow is clearly sequenced with a real input-validation checkpoint ("Check if necessary columns exist... at least 2 studies, HR and CI are positive numbers") and an explicit feedback loop ("If there are data issues, prompt the user to correct and resubmit"), plus a failure-handling section. Not a 5 because the documented post-implementation verification command (forest_survival.py --help) does not actually run as written, and the duplicated Validation Shortcut / Quick Validation sections create contradictory verification paths. | 4 / 5 |
Progressive Disclosure | The structure is flat and one level deep: SKILL.md points directly to scripts/forest_survival.py, scripts/forest_survival.R, and scripts/validate_skill.py, all of which exist in the bundle; no content that belongs in separate reference files is inlined. Minor gaps: "plus 1 additional script(s)" is inaccurate (there are 2 additional scripts), the cross-reference "See `## Workflow` above" points to a section that appears below it, and redundant sections (two validation sections, two dependency sections) hurt organization, keeping it below anchor 5. | 4 / 5 |
Total | 14 / 20 Passed |