Content
50%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 delivers genuinely useful operational detail — a complete parameter table, runnable examples, and real bundle references — but it is buried in a large amount of generic template boilerplate and meta-commentary that dilutes the signal. Cutting the padding and making the workflow section describe the actual search procedure would lift it substantially.
Suggestions
Cut the generic template sections (Security Checklist, Risk Assessment, Lifecycle Status, Evaluation Criteria, Response Template, Output Requirements) or move them to a references/ file — they add little FDA-search-specific value and roughly halve the token footprint.
Remove the 'See `## X` above for related details' pointers and merge the duplicated sections (When to Use vs Input Validation, Key Features vs Features, Example Usage vs Usage, Workflow vs Implementation Details) into single, concrete sections.
Rewrite the 'Workflow' section as the actual executable sequence (validate input → query/search → optional download → report results with assumptions) with the rate-limit and API-failure checkpoints inline, and fix the non-portable 'cd "20260318/..."' example path.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~260-line body is noticeably verbose with substantial template padding: a generic Security Checklist (12 unchecked boxes), Risk Assessment table, Lifecycle Status, Evaluation Criteria, Output Requirements, Error Handling, Input Validation, and Response Template sections that are boilerplate rather than FDA-search-specific knowledge, plus odd self-referential pointers ('See `## Features` above for related details', 'See `## Prerequisites` above') and a garbled trigger sentence ('the task needs Search FDA industry guidelines'). This matches 'Noticeably verbose; several unnecessary explanations or padded sections' rather than score 3's mostly-efficient baseline. | 2 / 5 |
Actionability | Concrete, executable guidance is present: a full parameter table with defaults and required flags, three copy-paste-ready example invocations ('python scripts/main.py --area oncology'), a Quick Check ('python -m py_compile scripts/main.py'), and a worked JSON output example. Minor gaps keep it below 5: the Example Usage 'cd "20260318/scientific-skills/..."' path is a non-portable build-artifact path, and the 'Example run plan' mixes executable steps with abstract direction. | 4 / 5 |
Workflow Clarity | A sequence exists and some validation checkpoints are present (py_compile quick check, --help inspection, Error Handling fallbacks), matching 'Steps listed but validation gaps'. However the numbered 'Workflow' section is generic governance prose ('Confirm the user objective...', 'switch to the fallback path') rather than the actual search workflow, and the network/ scraping path (rate-limited FDA API) lacks concrete retry/validation checkpoints, so it does not reach score 4's 'clear sequence with most checkpoints present'. | 3 / 5 |
Progressive Disclosure | The three referenced files (references/search-strategy.md, references/area-mappings.json, references/fda-api-notes.md) all exist in the bundle and are clearly listed one level deep in a References section, and scripts/main.py exists. But significant content that belongs in separate files is inlined (risk table, security checklist, lifecycle, response template), the 'See `## X` above' cross-references are confusing non-standard navigation, and the body's cache location ('references/cache/') does not exist in the bundle — matching 'Some structure but could be better organized' rather than score 4. | 3 / 5 |
Total | 12 / 20 Passed |