evo2

Score, embed, and generate DNA sequences with Evo 2, a long-context genomic foundation model. Use this skill when: (1) Computing per-nucleotide or per-sequence likelihoods for variant effect scoring, (2) Embedding genomic windows for downstream classification, (3) Generating DNA conditioned on a prefix, (4) Scoring regulatory or coding regions across species.

Quality

81%

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

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 tight, well-structured skill body with executable examples, a useful decision tree, and clear operational guidance for remote GPU compute. It sits just below top marks because a couple of code paths are described in prose and the workflow lacks an explicit output-validation checkpoint.

Suggestions

Provide the full `score_evo2.py` script (or a complete variant Δll code block) so the scoring path is fully copy-paste executable rather than partially described in prose.

Add an explicit output-validation checkpoint after `save_artifacts` (e.g. confirm `scores.json` is non-empty / well-formed before declaring success) to strengthen the batch-scoring workflow's feedback loop.

Consider moving the detailed Remote compute orchestration into a one-level-deep reference file to keep SKILL.md a lean overview and lift progressive disclosure.

Dimension	Reasoning	Score
Conciseness	Mostly lean with token-efficient tables and code blocks and no concept-padding, but the Remote compute prose (HF_HOME/HF_HUB_OFFLINE explanation, attach_job details) has minor trimmable instances keeping it just below 5.	4 / 5
Actionability	Copy-paste-ready code for loading, scoring, generating, and remote job submission covers common cases, but the referenced `score_evo2.py` script body is described in prose rather than given as code, and the Δll variant workflow is not a complete executable block.	4 / 5
Workflow Clarity	The remote-compute flow is clearly sequenced (read compute_details → submit_job → wait_for_notification → save_artifacts → attach_job) with a notification checkpoint and a troubleshooting error-recovery table, but lacks an explicit output-validation step in the main workflow.	4 / 5
Progressive Disclosure	Well-organized into clearly headed sections with no nested references and a self-contained structure, but at ~125 lines (above the simple-skill 50-line threshold) the dense Remote compute section could plausibly live in a separate reference file, leaving minor organization gaps.	4 / 5
	Total	16 / 20 Passed

Description

88%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 strong, specific description with explicit what/when structure and concrete enumerated use cases. Its main limitation is the absence of synonyms and file-format trigger terms that genomics users might naturally say.

Suggestions

Add natural synonyms and file-format triggers users say, e.g. "FASTA/.fasta files", "variants / SNVs / mutations", to lift trigger-term quality from 4 to 5.

Add a brief disambiguating clause to separate Evo 2 (likelihood/generation) from track-prediction models like borzoi, reducing conflict risk with sibling genomic skills.

Dimension	Reasoning	Score
Specificity	Four enumerated concrete capabilities ("Computing per-nucleotide or per-sequence likelihoods", "Embedding genomic windows", "Generating DNA conditioned on a prefix", "Scoring regulatory or coding regions") plus summary verbs give comprehensive, specific action coverage with no real gaps.	5 / 5
Completeness	Explicitly answers both "what" ("Score, embed, and generate DNA sequences with Evo 2") and "when" via four concrete enumerated trigger clauses under "Use this skill when:".	5 / 5
Trigger Term Quality	Strong natural domain terms ("DNA sequences", "variant effect", "embedding", "generate DNA", "regulatory/coding regions") but lacks synonyms and file-format triggers (FASTA, variants, SNVs, .fasta) that would push it to 5.	4 / 5
Distinctiveness Conflict Risk	A clear DNA-language-model niche (likelihood/embed/generation) with distinct triggers, but minor overlap risk with closely related genomic foundation models (e.g. borzoi, Nucleotide Transformer) that also score or embed DNA.	4 / 5
	Total	18 / 20 Passed

Validation

81%

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 — 13 / 16 Passed

Validation for skill structure

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

	Total	13 / 16 Passed

Repository: UnicomAI/wanwu
Path: configs/microservice/bff-service/configs/agent-skills/claude-science/evo2/SKILL.md
Commit: 030c20b

Reviewed: 19 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.