structure-prediction

Protein structure prediction from sequence. ESMFold-based, single GPU, no MSA needed. Predicts 3D structures with pLDDT confidence scores for drug discovery targets.

Quality

65%

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/structure-prediction/SKILL.md

Quality

Content

68%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 highly actionable with complete, executable commands for all four workflows and well-signaled one-level-deep references to real bundle files. Its weaknesses are duplicated pLDDT interpretation content (inline table vs. references/confidence_metrics.md), some over-explanation of ESMFold basics, and missing validation/verification steps for the batch workflow.

Suggestions

Trim the inline pLDDT score table and "What good/bad looks like" detail, pointing instead to references/confidence_metrics.md to remove duplication.

Add an explicit validation step for batch runs, e.g., "After batch prediction, check summary.csv for low pLDDT or failed entries and re-run failures with --device cpu if OOM occurred".

Shorten the Overview's explanation of what ESMFold is and its advantages, keeping only the details Claude would not already know (VRAM limits, speed tradeoffs).

Dimension	Reasoning	Score
Conciseness	The body is mostly efficient but the Overview explains what ESMFold is (a concept Claude already knows) and the detailed pLDDT interpretation table duplicates references/confidence_metrics.md, going beyond the "minor instances" of anchor 4 and fitting anchor 3.	3 / 5
Actionability	All four workflows provide fully executable, copy-paste-ready commands with real flags (--input, --output, --device, --output-dir) and concrete output descriptions, covering the common cases and matching anchor 5.	5 / 5
Workflow Clarity	The four workflows are clearly sequenced with inputs and outputs, but batch prediction (predict_batch.py) has no explicit validation/verification step or error-recovery loop, which caps workflow clarity at 3 per the batch-operations rule.	3 / 5
Progressive Disclosure	Structure is good: each workflow has a "See: scripts/..." pointer to a real file, and the confidence reference is one level deep and clearly signaled; the inline pLDDT table duplicating the reference content is a minor organization gap, matching anchor 4 rather than 5.	4 / 5
	Total	15 / 20 Passed

Description

61%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 concise and specific, clearly pinning the niche with "ESMFold-based, no MSA needed" and naming pLDDT outputs. Its main gap is the absence of an explicit "Use when..." trigger clause, which caps completeness, and limited natural trigger variations such as "fold" or "PDB".

Suggestions

Add an explicit 'Use when...' clause, e.g., "Use when predicting or folding protein structures from amino acid sequences, screening targets in batch, or checking pLDDT confidence."

Include natural trigger synonyms users would say, such as "fold protein", "3D structure", and "PDB", to broaden keyword coverage.

Mention the batch, evaluation, and comparison capabilities so the description covers all core workflows.

Dimension	Reasoning	Score
Specificity	The description names the domain and 1-2 concrete actions ("Predicts 3D structures with pLDDT confidence scores") but omits batch screening, evaluation, and comparison capabilities covered by the skill, matching anchor 3 rather than the more comprehensive anchor 4.	3 / 5
Completeness	The "what" is clear (ESMFold-based 3D structure prediction with pLDDT scores) but the "when" is only weakly implied by "for drug discovery targets" — there is no "Use when..." clause or equivalent explicit trigger guidance, capping completeness at 3.	3 / 5
Trigger Term Quality	Good keyword coverage including "protein structure prediction", "sequence", "ESMFold", "pLDDT", and "3D structures", but natural synonyms users would say such as "fold protein" or "PDB" are missing, placing it at anchor 4 rather than 5.	4 / 5
Distinctiveness Conflict Risk	"ESMFold-based, single GPU, no MSA needed" carves a clear niche versus MSA-based predictors, but there remains minor overlap risk with closely related protein structure prediction skills (e.g., AlphaFold-style), matching anchor 4.	4 / 5
	Total	14 / 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/structure-prediction/SKILL.md
Commit: 4082a2e

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.