Conciseness

The body is mostly efficient — command blocks, tables, and JSON examples all earn their tokens — but the MM/GBSA boundary disclaimer is repeated at least five times (Key capabilities, Validation Warning, 'MM/GBSA capability boundary', Workflow 2, 'Ligand Heuristic Scores', Troubleshooting), which is unnecessary padding. It is not 2 because no section explains concepts Claude already knows, and not 4 because the repetition is substantial rather than a minor trim.

Actionability

Every workflow is a complete, copy-paste-ready command with all flags shown (e.g. 'python scripts/predict.py --protein prepared_protein.pdb --poses docking_results/poses.sdf --output affinity.json'), plus install commands, quick verification snippets, a full 6-step pipeline, and concrete example outputs. Fully executable with the common cases covered.

Workflow Clarity

Workflows are clearly sequenced (Workflow 5 numbers steps 1-6) with a general verification convention — 'Verify the exit status and record status before using a score', the _script_manifest.jsonl check, failed-record handling, and a Troubleshooting section. It is not 5 because the batch screening workflow (Workflow 4) has no embedded validate-after-run checkpoint or fix-and-retry loop; it is not 3 because exit-status/manifest verification and troubleshooting do provide real checkpoints for the batch path.

Progressive Disclosure

Section structure is good (Overview → When to Use → Installation → Workflows → Script Reference → Output Format → Interpretation → Troubleshooting), but the bundled references/scoring_methods.md is never linked or mentioned anywhere in the body — a buried reference — and detailed content that belongs there (output interpretation tables, scoring-method detail) is inlined. It is not 2 because the body is well-sectioned rather than monolithic, and not 4 because an entire bundled reference file is undiscoverable from the entry point.

Specificity

The description lists several concrete, domain-specific actions — 'Empirical affinity estimates, ligand energy inspection, docking-score consensus, and batch virtual screening' — plus an explicit MM/GBSA boundary. It falls just below anchor 5 because the actions are terse noun phrases and omit key outputs (pKd/Kd/ΔG, pose ranking) that would make coverage comprehensive.

Completeness

The 'what' is clear (four named capabilities plus a boundary statement), but there is no 'Use when...' clause or equivalent explicit trigger guidance; the when is only weakly implied by the domain terms. Per the rubric guideline, a missing 'Use when' clause caps completeness at 3 — it is not 4 because the when is absent rather than merely imprecise.

Trigger Term Quality

Good natural keyword coverage: 'affinity', 'docking-score', 'consensus', 'virtual screening', 'MM/GBSA' are all phrases users would plausibly say. A few natural terms are missing — the literal phrases 'binding affinity', 'predict', 'Kd', 'pKd' never appear, so it sits at anchor 4 rather than 5.

Distinctiveness Conflict Risk

It carves a clear niche — post-docking affinity estimation — and explicitly disclaims full MM/GBSA ('Full MM/GBSA requires a validated external workflow'), minimizing conflict with molecular-docking, pocket-detection, or MD skills. It is not 4 because the only overlap ('docking-score') is immediately qualified as consensus scoring of docking results, not pose generation.