torchdrug

PyTorch-native graph neural networks for molecules and proteins. Use when building custom GNN architectures for drug discovery, protein modeling, or knowledge graph reasoning. Best for custom model development, protein property prediction, retrosynthesis. For pre-trained models and diverse featurizers use deepchem; for benchmark datasets use pytdc.

Quality

79%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./backend/cli/skills/chemistry/torchdrug/SKILL.md

Quality

Content

71%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.

The body is a well-organized overview with executable code, clear workflows, and excellent progressive disclosure to real reference files. Its main weakness is redundancy: reference pointers and capability summaries are repeated across multiple sections, inflating the token budget without adding information.

Suggestions

Consolidate the reference pointers into a single navigation block (e.g., keep the Summary or Cheat Sheet and remove the duplicated 'See references/X.md for:' lists from Core Capabilities).

Add explicit validation/feedback checkpoints inside the Common Workflows (e.g., 'verify loss decreases and check metric on validation split before proceeding').

Tighten the Overview paragraph and Core Capabilities sections to avoid restating task lists that already appear in 'When to Use This Skill'.

Dimension	Reasoning	Score
Conciseness	Mostly efficient and well-structured, but the eight reference pointers are repeated three times (Core Capabilities, Quick Reference Cheat Sheet, and Summary) and the Common Workflows restate capability content, adding padding that could be trimmed.	3 / 5
Actionability	The Quick Example and Integration Patterns provide copy-paste-ready executable code with real imports and parameters, but the Common Workflows are high-level numbered step lists without full executable code for each step.	4 / 5
Workflow Clarity	Five workflows are clearly sequenced with numbered steps and end-of-pipeline evaluation checkpoints (e.g., 'Evaluate using AUROC and AUPRC'), though intermediate validation/feedback loops are mostly implicit rather than explicit.	4 / 5
Progressive Disclosure	A concise overview points to eight real, one-level-deep reference files (all present in references/), each clearly signaled with the topics it covers, plus a cheat sheet and summary for easy navigation.	5 / 5
	Total	16 / 20 Passed

Description

87%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.

The description is strong: it states what TorchDrug does, when to use it, and how it differs from deepchem and pytdc. Trigger phrases are concrete and natural. Minor gaps in action specificity and a few missing synonyms keep specificity and trigger_term_quality just below the top anchor.

Dimension	Reasoning	Score
Specificity	Lists several concrete tasks ('building custom GNN architectures', 'protein modeling', 'knowledge graph reasoning', 'protein property prediction', 'retrosynthesis') rather than vague language, though they are domain-level rather than fully concrete actions.	4 / 5
Completeness	Explicitly answers both what ('PyTorch-native graph neural networks for molecules and proteins') and when ('Use when building custom GNN architectures for drug discovery, protein modeling, or knowledge graph reasoning') with concrete trigger phrases.	5 / 5
Trigger Term Quality	Natural user-facing terms like 'drug discovery', 'protein modeling', 'knowledge graph reasoning', and 'retrosynthesis' appear alongside library names (deepchem, pytdc), but common variations such as SMILES or molecular generation are absent.	4 / 5
Distinctiveness Conflict Risk	Carves a clear niche and actively disambiguates from adjacent tools ('For pre-trained models and diverse featurizers use deepchem; for benchmark datasets use pytdc'), minimizing conflict risk.	5 / 5
	Total	18 / 20 Passed

Validation

87%

Warnings & errors only

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

Validation — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/chemistry/torchdrug/SKILL.md
Commit: 3a6c3a9

Reviewed: 1 day ago

Table of Contents

Discovery Implementation Validation

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.