CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

diffdock

Diffusion-based molecular docking. Predict protein-ligand binding poses from PDB/SMILES, confidence scores, virtual screening, for structure-based drug design. Not for affinity prediction.

Quality

77%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./scientific-skills/diffdock/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

75%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A strong, highly actionable tool skill: every workflow is backed by executable commands and the batch pipeline includes genuine validation checkpoints. The weaknesses are verbosity — redundant Resources/Best Practices sections, marketing language, and an off-topic product plug — and reference-content duplication that blurs the overview-vs-reference split.

Suggestions

Trim the Resources section to a compact table of file + one-line purpose, since each script/reference is already introduced where it is used.

Replace the inline confidence table and parameter-tuning details with pointers to confidence_and_limitations.md and parameters_reference.md, keeping only the decision-critical summary.

Add an explicit feedback loop for batch runs: what to do when `prepare_batch_csv.py --validate` reports errors (fix rows, re-validate, then run).

DimensionReasoningScore

Conciseness

The core sections (workflows, commands, parameters, troubleshooting) are dense and earn their tokens, but there is real padding: "represents the state-of-the-art... crucial for" marketing language, a ~60-line Resources section that re-describes scripts/references already introduced inline, Best Practices items repeating earlier guidance, and an off-topic K-Dense Web promotion. Not a 2 because the padding is a minority (~25%) of otherwise tight, executable content.

3 / 5

Actionability

Fully executable, copy-paste-ready commands throughout: exact `python -m inference` invocations with real flags, the batch CSV format with a worked example, analyze_results.py usage with flags, and a runnable pandas snippet for ensemble docking. The common cases (single docking, batch/screening, result analysis) are each covered by concrete runnable examples.

5 / 5

Workflow Clarity

The three workflows are clearly sequenced, and the batch path has real validation checkpoints (setup_check.py before jobs, `prepare_batch_csv.py --validate` before batch runs, confidence-thresholded analysis after). It falls short of 5 because the validate-fail-fix-retry feedback loop is implicit — the body never says what to do when CSV validation fails — and the batch workflow relies on the reader connecting troubleshooting entries to earlier steps.

4 / 5

Progressive Disclosure

Good structure with three real, one-level-deep reference files (verified present, no nested pointers), each clearly signaled ("Read `references/...` using the Read tool") plus a Resources section explaining when to read each. Not a 5 because the 475-line body inlines material that duplicates the references (confidence table, parameter tuning, and limitations sections all repeat content in confidence_and_limitations.md and parameters_reference.md) rather than pointing to them.

4 / 5

Total

16

/

20

Passed

Description

70%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A crisp, third-person description that names concrete capabilities, input formats, and an explicit out-of-scope boundary. Its main weakness is the absence of a 'Use when...' trigger clause, which caps completeness and limits natural trigger coverage.

Suggestions

Append an explicit trigger clause, e.g. "Use when asked to dock ligands, predict binding poses, or run virtual screening from PDB or SMILES inputs."

Add common user phrasings as trigger terms ("dock this ligand", "binding site", "screen a compound library") to broaden natural-language matching.

Mention sequence input via ESMFold to make the description's capability coverage match the body.

DimensionReasoningScore

Specificity

The description lists several concrete actions — "Predict protein-ligand binding poses from PDB/SMILES", "confidence scores", "virtual screening" — with concrete input formats (PDB/SMILES). It falls short of a 5 because coverage has minor gaps (no mention of sequence/ESMFold input or batch processing, both core capabilities documented in the body).

4 / 5

Completeness

The 'what' is clearly answered (pose prediction, confidence scores, virtual screening), but there is no explicit 'Use when...' clause or equivalent trigger guidance — "for structure-based drug design" only weakly implies when. Per the rubric guideline, a missing 'Use when' clause caps completeness at 3; this is not a 4 because 'when' is neither explicit nor concrete, and not a 2 because the 'what' is fully clear.

3 / 5

Trigger Term Quality

Strong natural terms users would actually say: "molecular docking", "binding poses", "virtual screening", "PDB/SMILES", "structure-based drug design". Not a 5 because a few common variations are missing (e.g. "dock this ligand", "binding site", "screen compound library", "pose prediction").

4 / 5

Distinctiveness Conflict Risk

"Diffusion-based molecular docking" carves out a clear niche, and the explicit negative boundary "Not for affinity prediction" actively disambiguates from scoring-function/affinity skills. Minimal conflict risk; matches the clear-niche anchor exactly.

5 / 5

Total

16

/

20

Passed

Validation

93%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

metadata_version

'metadata.version' is missing

Warning

Total

15

/

16

Passed

Repository

googolme/run0204

Path

scientific-skills/diffdock/SKILL.md

Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.