Content
78%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.
A highly actionable body with executable commands and genuine validation gating for batch pipeline runs, weakened by dense inline output inventories that belong in reference files and some changelog-flavored prose. Restructuring the per-backend details into one-level-deep references would address its main weakness.
Suggestions
Split the backend output inventories (tables/, visualizations/, resources/ file lists) into a reference file such as references/outputs.md and link to it from the main body.
Trim changelog-style phrasing like 'The runner now also emits...' into plain present-tense statements of what each command produces.
Add an explicit 'if validation fails, fix and re-run' step after the preflight and qc_verdict checks to close the feedback-loop gap.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dominated by concrete commands, flags, and file paths with zero explanation of concepts Claude already knows. However, changelog-style sentences like 'The runner now also emits methods/amplicon_methods.json plus a concrete backend handoff bundle...' and the long multi-clause paragraphs on outputs (lines 43, 64, 91, 93) could be tightened or trimmed. | 4 / 5 |
Actionability | Multiple copy-paste-ready bash commands with concrete arguments (real primer sequences, a named classifier file, specific backends, explicit flags and output paths), covering preflight, FASTQ QC, QIIME2, DADA2, and nf-core paths. The common cases are covered with executable commands. | 5 / 5 |
Workflow Clarity | A clear progression exists (essential inputs → preflight → FASTQ QC package → backend execution → visualization) with real validation checkpoints: preflight install plan, readiness verdicts, qc_verdict.json, gating that 'FASTQs alone' are not analysis-ready, and the review-only block on synthetic diversity data. It falls short of a 5 because there is no explicit fix-and-retry feedback loop for failed validation and the batch-run sequence is implied by section order rather than enumerated steps. | 4 / 5 |
Progressive Disclosure | Sections are well organized, but no bundle files exist (no references/, scripts/, or assets/), so dense inline material — the backend output-file inventory (lines 66-75), the resource-plan flag detail (line 64), and the visualization output catalog (line 93) — is content that could live in one-level-deep reference files. At ~95 lines of detail across three backends, the inlining is more than a minor organization gap, matching the 'content that should be separate is inline' anchor. | 3 / 5 |
Total | 16 / 20 Passed |