Content
46%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 workable ROB2 workflow with a proper one-level reference file and real scripts, but it is padded with auto-generated boilerplate sections, presents four overlapping run descriptions, and includes command examples that do not execute as written. It reads as an unedited generated skill rather than a curated one.
Suggestions
Remove the boilerplate sections (Key Features, Dependencies, Implementation Details) and the near-verbatim restatement of the description in 'When to Use'; they consume tokens without adding actionable guidance.
Fix the executable examples: `scripts/assess_rob2.py` has no CLI so `--help` fails, and `python extract_pdf.py` needs the PDF path argument and the `scripts/` prefix — show the actual invocations from the script's own docstring.
Consolidate the four overlapping run descriptions (## Usage, ### Workflow Steps, Example run plan, Usage flow) into one sequenced workflow, and correct the 'See `## Usage` above' cross-reference, which points to a section that appears later.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Several padded boilerplate sections ("Scope-focused workflow aligned to:", "Structured execution path designed to keep outputs consistent and reviewable", "Execution model: validate the request, choose the packaged workflow, and produce a bounded deliverable") plus a When to Use section restating the description nearly verbatim with garbled grammar ("use when you need automates risk of bias 2 (rob2) assessment..."). This matches anchor 2's several unnecessary/padded sections rather than anchor 3's occasional tightening needs. | 2 / 5 |
Actionability | There is concrete guidance (the JSON output schema, the `from scripts.assess_rob2 import clean_text` example), but commands are broken as written: `python scripts/assess_rob2.py --help` fails because the script defines functions with no CLI, `python extract_pdf.py` omits the required PDF-path argument and its `scripts/` location, and "See `## Usage` above" points to a section that appears later in the document. The gaps exceed anchor 4's 'minor gaps'. | 3 / 5 |
Workflow Clarity | A clear sequence exists (Extract Study Info → Assess Domains → Synthesize, and the PDF extract → read → assess flow), but validation checkpoints are absent or implicit and four overlapping run descriptions (## Usage, ### Workflow Steps, Example run plan, Usage flow) blur which is canonical — matching anchor 3's 'steps listed but checkpoints missing'. | 3 / 5 |
Progressive Disclosure | The one-level reference is clearly signaled ("Refer to [rob2_guidelines.md](references/rob2_guidelines.md) for the detailed questions and logic for each domain" — verified to exist) and the two scripts are real and referenced. Minor gaps: domain synthesis rules are duplicated inline from the reference, and `extract_pdf.py` is referenced without its `scripts/` path, so it falls just short of anchor 5. | 4 / 5 |
Total | 12 / 20 Passed |