Content
38%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 skill has a genuinely useful core (arguments, input/output formats, statistical methods, example results) buried in extensive generic template boilerplate and broken structural references. It reads as an auto-filled template rather than curated guidance, with missing bundle files and duplicated/malformed parameter documentation.
Suggestions
Delete the generic template sections (Risk Assessment, Security Checklist, Evaluation Criteria, Lifecycle Status, Response Template, Output Contract, Inputs to Collect, Validation and Safety Rules) — none contain survival-analysis-specific information Claude cannot infer.
Fix the broken references: remove the bogus 'cd "20260318/..."' path, correct the 'See ## Features/Usage/Workflow above' pointers (those sections appear later), and either ship the referenced scripts/main.py, references/, and requirements.txt or stop citing them.
Merge the malformed duplicate 'Parameters' table (empty descriptions for --event/--group, a broken '--figsize | str | 10' default, and --risk-table inconsistent between Required and Optional) into the single accurate 'Arguments' table, and make the example invocations uncommented, runnable commands.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Roughly half of the ~320-line body is generic template boilerplate ('Risk Assessment', 'Security Checklist', 'Evaluation Criteria', 'Lifecycle Status', 'Response Template', 'Output Contract', 'Validation and Safety Rules') that adds nothing specific to survival analysis, plus a duplicated and malformed 'Parameters' table alongside the earlier 'Arguments' table. It is not 1 because the core usage content (Arguments, Input Format, Output Files, Statistical Methods) is itself tight and specific rather than tutorial-style padding. | 2 / 5 |
Actionability | Concrete guidance exists (full argument table, example CSV input format, named output files, 'python -m py_compile scripts/main.py'), but all example invocations are commented out ('# Example invocation: python scripts/main.py ...') rather than executable, and the referenced artifacts (scripts/main.py, references/, requirements.txt) do not exist in the bundle, with a bogus hardcoded 'cd "20260318/..."' path. This incompleteness goes beyond the 'minor gaps' of a 4. | 3 / 5 |
Workflow Clarity | A sequence is present (confirm inputs -> py_compile check -> run scripts/main.py -> review outputs) with an explicit validation checkpoint and error-handling/fallback guidance, but the actual 'Workflow' and 'Example run plan' sections are generic boilerplate ('Validate that the request matches the documented scope'), duplicated across multiple sections, and never concretely tied to the survival-analysis steps. Not 4 because the checkpoints are implicit and scattered rather than forming one clear task-specific sequence. | 3 / 5 |
Progressive Disclosure | The body is a monolithic document with content that belongs in separate files inlined (security checklists, evaluation criteria, lifecycle status), and its file references are broken: 'scripts/main.py', 'references/', and 'requirements.txt' are cited but absent from the bundle, while pointers like 'See `## Features` above' and 'See `## Usage` above' reference sections that actually appear later in the document. It is above 1 only because the many section headers do provide some navigation. | 2 / 5 |
Total | 10 / 20 Passed |