Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
Highly actionable and clearly sequenced with validation checkpoints and fallbacks, but the body is long and bundles clinical teaching that could be tightened or moved into referenced files to improve token efficiency and progressive disclosure.
Suggestions
Trim the clinical teaching sections (histological vs molecular classification, staging vs grading, biomarker interpretation strategy) to the tool-use guidance Claude needs, or move them into a referenced reference file to improve conciseness.
Add a references/ bundle file (e.g., BIOMARKERS.md or EVIDENCE.md) and link to it one level deep to offload the dense interpretation material and raise progressive disclosure.
Move the verified codes table and biomarker examples closer to the relevant workflow phases so the core tool-use loop stays compact.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient with tool tables and terse code blocks, but the ~200-line body teaches substantial clinical oncology (histological vs molecular classification, staging vs grading, biomarker interpretation, evidence grading) that pads beyond what Claude needs to drive the tools. | 2 / 3 |
Actionability | Fully executable tool calls with real parameters and response-field annotations ('OncoTree_search(query="breast cancer")', 'OncoKB_annotate_variant(gene="EGFR", variant="L858R", tumor_type="LUAD")') plus a verified codes table — copy-paste ready. | 3 / 3 |
Workflow Clarity | A clear 4-phase sequence with an explicit validation checkpoint ('Always validate via OncoTree_get_type before using in downstream tools') and fallback chains for 404/unrecognized-code recovery. | 3 / 3 |
Progressive Disclosure | Well-sectioned single file, but no bundle files exist and the dense biomarker/evidence-grading material that could be offloaded is all inline rather than split into one-level-deep references. | 2 / 3 |
Total | 10 / 12 Passed |