Content
42%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 contains a serviceable workflow, error-handling, and response-template core, but is buried under generic template boilerplate and undermined by inaccurate references (missing requirements.txt, nonexistent CONFIG block, dated example path, forward self-references). The actual analysis procedure — input format, CLI flags, and how to interpret output — is never concretely documented.
Suggestions
Document the real script usage: input CSV format (time,concentration columns), the --data and --demo flags, and expected output parameters; remove the 'in-file CONFIG block' instruction, which does not match scripts/main.py.
Cut the boilerplate sections (Risk Assessment, Security Checklist, Evaluation Criteria, Lifecycle Status) and merge the duplicated Quick Check / Audit-Ready Commands blocks; they repeat the same two commands and add token cost without operational value.
Fix broken references: the missing requirements.txt (either add it or drop the dependency declarations), the non-reproducible '20260318/scientific-skills/...' example path, and the forward-pointing 'See ## Prerequisites/Workflow above' lines that appear before those sections.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is noticeably padded: boilerplate sections (Risk Assessment table, Security Checklist, Evaluation Criteria, Lifecycle Status with a 'Next Review Date' of 2026-03-06) add nothing Claude needs, the same py_compile/--help commands are repeated across 'Quick Check' and 'Audit-Ready Commands', and 'When to Use' restates the frontmatter description verbatim twice. This matches anchor 2 ('noticeably verbose; several unnecessary explanations or padded sections'); not 1 because genuine operational content (Parameters, Returns, Workflow, Error Handling) is also present. | 2 / 5 |
Actionability | Some concrete commands exist and the referenced script is real ('python -m py_compile scripts/main.py', 'python scripts/main.py --help'), but the key execution details are missing: the input file format (CSV with time,concentration columns), the actual CLI flags (--data, --demo), and the 'Example Usage' path ('20260318/scientific-skills/...') is not reproducible. The instruction to 'Edit the in-file CONFIG block' points to a block that does not exist in scripts/main.py. This matches anchor 3 ('some concrete guidance but incomplete; missing key details'); not 4 because a user could not run a real analysis from the body alone. | 3 / 5 |
Workflow Clarity | The Workflow section lists a clear 5-step sequence with scope validation ('stop early if the task would require unsupported assumptions') and a fallback path ('If execution fails or inputs are incomplete, switch to the fallback path'), plus a pre-execution parse check. However, there is no validation of computed results, the steps are process-general rather than PK-specific, and the section cross-references ('See ## Workflow above', 'See ## Prerequisites above') appear before the sections they cite. This matches anchor 3 ('sequence present but checkpoints missing or implicit'); not 4 because the verification gaps and broken references undermine the checkpoints. | 3 / 5 |
Progressive Disclosure | Structure exists with clear section headers and a real, correctly referenced script (scripts/main.py), but organization is loose: 'requirements.txt' is cited as declaring dependencies yet does not exist in the bundle, self-references point the wrong direction, and a large amount of boilerplate (risk tables, checklists, lifecycle metadata) is inlined that belongs in separate governance files or nowhere. This matches anchor 3 ('some structure but could be better organized; references present but not clearly signaled'); not 2 because the core script reference is real and navigation is possible. | 3 / 5 |
Total | 11 / 20 Passed |