foldseek-structural-search

Performs 3D structural searches of proteins against various databases (PDB, AlphaFold, CATH, MGnify, etc.) using the Foldseek API. Use ONLY when the user provides a physical 3D coordinate file (.cif, .mmcif, or .pdb) and wants to find structurally similar proteins. Do NOT use if the user only provides a protein sequence, gene name, or UniProt ID.

Quality

75%

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 ./skills/foldseek_structural_search/SKILL.md

Quality

Content

63%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 content is actionable and sequenced with explicit validation gates, but it is held back by redundancy between the Core Rules and Instructions sections, an unreferenced bundle file (citation.bib), and a malformed orphaned heading for error handling. Tightening the repetition and integrating the error-recovery guidance would lift the weaker dimensions.

Suggestions

Consolidate the file-requirement rule so it is stated once (e.g., in Instructions step 1) and remove its duplication in Core Rules and the Goal section to improve conciseness.

Fix the malformed trailing heading '## * If the API returns an error...' by converting it into a proper numbered step (e.g., '10. Error handling:') so the error-recovery feedback loop is part of the workflow rather than orphaned prose.

Either reference references/citation.bib from the body (e.g., in the Goal or a Citations section) or remove it from the bundle so the reference set is clearly signaled and consistent with progressive disclosure.

Dimension	Reasoning	Score
Conciseness	Mostly efficient with useful domain-specific metric explanations (Prob, Q-Cov, E-value), but the file-requirement point is restated three times (Core Rules 'File Requirement', Goal, and Instructions step 1) and the Core Rules section largely duplicates the Instructions, so it could be tightened — matching the mostly-efficient-but-could-be-tighter anchor rather than the lean anchor above.	3 / 5
Actionability	Provides concrete executable commands ('uv run scripts/search.py <path-to-file> -o <generated-filename.json> > <generated-filename.md>') for both default and custom-database cases plus an explicit database allowlist, which is mostly copy-paste ready; the parameterized placeholders and a slightly under-specified redirect keep it just below the fully-executable anchor.	4 / 5
Workflow Clarity	A clear 9-step sequence with explicit validation gates ('halt immediately' for missing coordinate file and for unsupported databases) and an error-recovery note, satisfying the clear-sequence-most-checkpoints anchor; it falls short of 5 because the error-handling guidance is malformed (an orphaned heading '## * If the API returns an error...') rather than an integrated feedback loop.	4 / 5
Progressive Disclosure	The body is organized into clear sections (Prerequisites, Goal, Core Rules, Instructions) and references scripts/search.py via inline commands, but references/citation.bib is a bundle file never linked or signaled from the body and the trailing orphaned error-handling heading breaks the structure, matching the some-structure-but-could-be-better-organized anchor.	3 / 5
	Total	14 / 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 third-person, well-scoped, and clearly answers both what the skill does and when to use it, with strong negative-trigger guidance that minimizes conflict risk. It is slightly below top marks only on specificity and trigger-term breadth, since the capability is a single richly-described action rather than a list of several.

Dimension	Reasoning	Score
Specificity	Names the domain and concrete action ('Performs 3D structural searches of proteins against various databases (PDB, AlphaFold, CATH, MGnify, etc.) using the Foldseek API') with specific databases and file types, but the capability is essentially one action richly elaborated rather than multiple distinct actions, so it sits just below the comprehensive-coverage anchor.	4 / 5
Completeness	Explicitly answers 'what' (Performs 3D structural searches... using the Foldseek API) and 'when' with a concrete trigger ('Use ONLY when the user provides a physical 3D coordinate file (.cif, .mmcif, or .pdb) and wants to find structurally similar proteins'), matching the clearly-and-explicitly-both-what-and-when anchor.	5 / 5
Trigger Term Quality	Includes natural domain phrases ('3D structural searches of proteins', 'find structurally similar proteins') plus file extensions (.cif, .mmcif, .pdb) and contrast terms (protein sequence, gene name, UniProt ID); good coverage though a few lay synonyms a biologist might say are absent, placing it just below the comprehensive anchor.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (3D coordinate-file structural search via Foldseek) and the explicit negative guidance ('Do NOT use if the user only provides a protein sequence, gene name, or UniProt ID') sharply reduces overlap with sequence-based skills, matching the clear-niche-minimal-conflict anchor.	5 / 5
	Total	18 / 20 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: google-deepmind/science-skills
Path: skills/foldseek_structural_search/SKILL.md
Commit: 6883275

Reviewed: 4 days 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.