Content
50%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-structured overview of a complex 9-phase pipeline with clear sequencing and concrete tool/parameter tables, but it relies on reference files that are absent from the bundle and lacks in-body validation checkpoints and runnable examples. Moderate quality with structural gaps to close.
Suggestions
Ship the referenced bundle files (phase-procedures.md, report-template.md, reference-data.md, tool-reference.md) or inline the essential per-phase steps so the signaled references are not dead links.
Add explicit per-phase validation checkpoints (e.g., confirm gene-ID resolution succeeded and enrichment FDR<0.05 filter applied before advancing) rather than relying on the end-of-run checklist alone.
Provide at least one runnable Python example for the 'COMPUTE, DON'T DESCRIBE' directive and trim the redundant 'Summary' section that restates the phase list.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly operational (tables, tool lists, scoring) with little concept explanation, but the closing 'Summary' section restates the phase list and 'Common Use Cases' adds length that could be tightened. | 2 / 3 |
Actionability | Names specific tools, parameters, FDR<0.05 thresholds, and a report template, but the executable per-phase workflows are deferred to reference files and no runnable code example is given for the 'COMPUTE, DON'T DESCRIBE' directive. | 2 / 3 |
Workflow Clarity | Phases 0-8 are clearly numbered and sequenced with a completeness checklist, but per-phase validation/feedback checkpoints are implicit or deferred rather than explicit in the body. | 2 / 3 |
Progressive Disclosure | The body is structured as an overview with clearly signaled one-level-deep references (phase-procedures.md, report-template.md, etc.), but none of those bundle files actually exist in the repo, so navigation is broken. | 2 / 3 |
Total | 8 / 12 Passed |