tooluniverse-variant-to-mechanism

End-to-end variant-to-mechanism analysis — trace a variant (rsID/coordinates) through regulatory context, target gene(s), molecular pathway(s), and phenotypic consequences. Integrates 7+ databases across 3 evidence layers (regulatory, molecular, disease) for a mechanistic model. Use for GWAS-hit-to-mechanism, eQTL-causal-gene tracing, and full causal-chain reports.

Quality

86%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

85%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.

An operationally excellent skill: executable code with correct parameter names, response formats, decision thresholds, explicit fallbacks for missing data, and a well-sequenced six-phase workflow with evidence grading. Its main structural weakness is that it is a single monolithic file — the per-tool API details belong in a reference file with SKILL.md kept as the overview and workflow guide.

Suggestions

Move the per-tool parameter documentation and response-format details (Phases 1-5 code blocks, ~150 lines) into a references/tool-api.md file, keeping SKILL.md as the phase overview with one or two key calls per phase and a clearly signaled link (e.g., "See [references/tool-api.md](references/tool-api.md) for all tool parameters and response formats").

Deduplicate the parameter pitfalls: state each gotcha once (either inline in the code comment or in "Common Mistakes and Fallbacks", not both) — the MONDO-vs-EFO rule and several param-name warnings currently appear two to three times.

Extract the tool-by-tool notes (e.g., "GTEx uses v8 data (v10 endpoints may return empty)", "gwas_get_associations_for_trait is BROKEN") into a compact reference table so the workflow sections carry only phase-level reasoning.

Dimension	Reasoning	Score
Conciseness	Code blocks are dense with tool-specific gotchas, return shapes, and thresholds that Claude would not know, so most tokens earn their place; however, pitfalls are stated redundantly (MONDO-not-EFO appears in section 2e, Phase 5, and Common Mistakes; param-name warnings appear both inline and in "Common Mistakes"), which is a minor trim opportunity. Not 5 because of that repetition; not 3 because the bulk is non-obvious operational knowledge rather than padding.	4 / 5
Actionability	Fully executable ToolUniverse calls with correct parameter names ("variant_id, NOT rsid"), expected response formats, decision thresholds (CADD >=20, L2G > 0.5, RegulomeDB 1a-2a), and a working REST fallback with pandas aggregation — copy-paste ready and covering the common cases. Matches the top anchor.	5 / 5
Workflow Clarity	Six phases are explicitly sequenced with an overview map, and error recovery is built in: "Response format variable: list, {data, metadata}, or {error} -- handle all three", a "When data is missing, adapt the analysis" fallback table, broken-tool warnings, and evidence-graded synthesis with confidence tiers. The workflow is read-only analysis, so the destructive/batch validation cap does not apply; the fallback and error-handling guidance constitutes the feedback loops the top anchor requires.	5 / 5
Progressive Disclosure	No bundle files exist (references/, scripts/, assets/ are absent) and the entire 370-line skill is monolithic: roughly 150 lines of per-tool parameter documentation (Phases 1-5 code blocks) are reference material inlined in SKILL.md, and no external reference is signaled (the lone "See tooluniverse-data-wrangling skill" mention points to a sibling skill, not a bundle file). Section structure is clear, which keeps it above 2, but "content that should be separate is inline" fits anchor 3 better than 4, where references would be mostly present and clear.	3 / 5
	Total	17 / 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 description: it states a concrete multi-step capability in third person, quantifies scope (7+ databases, 3 evidence layers), and includes an explicit "Use for" clause with domain-natural trigger phrases. The only weaknesses are a few missing common user phrasings (e.g., "how does this variant cause X") and no sibling-skill boundary statements in the description itself.

Dimension	Reasoning	Score
Specificity	"trace a variant (rsID/coordinates) through regulatory context, target gene(s), molecular pathway(s), and phenotypic consequences" lists multiple concrete actions covering the full analysis chain in third person, matching the comprehensive-coverage anchor; not 4 because there are no meaningful gaps in the action list.	5 / 5
Completeness	Explicitly answers both "what" (trace a variant through regulatory context, target genes, pathways, phenotypic consequences) and "when" ("Use for GWAS-hit-to-mechanism, eQTL-causal-gene tracing, and full causal-chain reports") with concrete trigger phrases, matching the top anchor; the "when" clause is explicit, not weakly implied.	5 / 5
Trigger Term Quality	"Use for GWAS-hit-to-mechanism, eQTL-causal-gene tracing, and full causal-chain reports" plus "rsID" and "mechanistic model" give good natural-keyword coverage, but common user phrasings like "how does this variant cause..." or plain "target gene" are missing; not 3 because several natural domain terms users would actually say are present.	4 / 5
Distinctiveness Conflict Risk	"variant-to-mechanism", "eQTL-causal-gene tracing" define a clear niche with distinct triggers, but the description omits the boundary guidance against closely related sibling skills (variant-interpretation, gwas-snp-interpretation) that the body contains, leaving minor overlap risk in the GWAS-variant space; not 5 for that missing boundary, not 3 because the niche is genuinely specific.	4 / 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: mims-harvard/ToolUniverse
Path: plugins/tooluniverse/skills/tooluniverse-variant-to-mechanism/SKILL.md
Commit: 164aaf1

Reviewed: 6 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.