CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

protein-sequence-similarity-search

Searches for homologous protein sequences using MMseqs2 (fast, default) or BLAST (comprehensive, fallback). Trigger this whenever the user provides a protein sequence or FASTA file and asks to find homologues, sequence matches, or wants to infer protein function based on sequence similarity, but not when the user wants to infer protein function based on structural similarity.

Quality

82%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Low

Low-risk findings worth noting

SKILL.md
Quality
Evals
Security

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 delivers fully executable commands with a clear dual-path workflow and genuine validation checkpoints (exit-code fallback, database whitelist). It is held back by padding in the database list and metric explanations, an unreferenced bundled citation.bib, and a couple of unhandled failure branches.

Suggestions

Move the Available BLAST Databases list to a references file (e.g., references/blast_databases.md) with a one-line validation rule, keeping only the most common codes inline.

Trim the per-database descriptions to a single short phrase and drop the metric glosses for E-value/Q-Cov/Seq Identity, which restate knowledge Claude already has.

Reference references/citation.bib explicitly (e.g., when to cite MMseqs2/ColabFold or EBI BLAST in output) and specify the behavior when BLAST also fails or the .md file is not produced.

DimensionReasoningScore

Conciseness

The core instructions are efficient, but the ~60-line database list redundantly restates every entry (e.g., "uniprotkb_swissprot – UniProtKB/Swiss-Prot (The manually annotated section of UniProtKB): The manually curated subsection...") and the metric glosses explain E-value/sequence-identity concepts Claude already knows. It could be tightened by roughly half without losing operational content.

3 / 5

Actionability

Copy-paste-ready commands ("uv run scripts/mmseqs2_search.py <SEQUENCE_OR_FILE> -o <generated-filename.md> -j <generated-filename.json>") cover the default, mgnify, and custom-database cases, with concrete filename guidance, an exit-code-2 fallback rule, and an explicit database-validation policy.

5 / 5

Workflow Clarity

The sequence (prerequisites → method selection → Path A/B → common interpretation) is clear with real checkpoints: the exit-code-2 fallback loop and the "do not proceed" database validation. Minor gaps keep it below anchor 5: no handling defined if BLAST also fails, no check that the .md output was generated before reading, and a grammar glitch ("of if you are unsure") muddies the method-selection rule.

4 / 5

Progressive Disclosure

Both bundled scripts are real, well-signaled, and integrated with clear usage, but references/citation.bib is shipped and never mentioned anywhere in the body, and the long database list is inline reference data that could live in a references file. Structure exists, but a bundled reference is buried and content that belongs in a separate file is inlined.

3 / 5

Total

15

/

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 a strong, explicit what+when statement with concrete natural-language triggers and a well-drawn boundary against structural-similarity requests. Its only weakness is that it describes a single capability area and omits a few natural keyword variants (e.g., ".fasta", "homologs").

DimensionReasoningScore

Specificity

"Searches for homologous protein sequences using MMseqs2 (fast, default) or BLAST (comprehensive, fallback)" names a concrete action, both tools, and their roles. It stays at one capability area rather than the several distinct actions of anchor 5, but is well above the generic 1-2-action anchor 3.

4 / 5

Completeness

The first sentence explicitly states what the skill does, and "Trigger this whenever the user provides a protein sequence or FASTA file and asks to find homologues..." gives explicit when-guidance with concrete trigger phrases, matching the anchor 5 example pattern. It even adds a negative boundary for structural similarity.

5 / 5

Trigger Term Quality

Natural triggers include "protein sequence", "FASTA file", "find homologues", "sequence matches", and "sequence similarity", with both tool names present. A few natural variations are missing (".fasta" extension, US spelling "homologs", "similar sequences"), so it falls just short of anchor 5's comprehensive synonym coverage.

4 / 5

Distinctiveness Conflict Risk

It occupies a clear niche (sequence-similarity homologue search via MMseqs2/BLAST) with tool-specific triggers, and the exclusion "but not when the user wants to infer protein function based on structural similarity" minimizes overlap with structure-based skills.

5 / 5

Total

18

/

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

google-deepmind/science-skills

Path

skills/protein_sequence_similarity_search/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.