CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

pubmed-database

Search PubMed for scientific literature, including published clinical trials. Fetch abstracts and full text. Link published research to biological databases (gene, protein, nucleotide, PubChem) to discover associations between papers and specific compounds or genes. Verify medical spelling, match raw citations, and cache result sets for bulk processing. Interfaces NCBI E-utilities and PMC BioC APIs.

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

82%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 well-structured, highly actionable skill body with effective progressive disclosure to real, one-level-deep reference files. The main improvements are removing duplicated bulk-workflow guidance and closing minor validation gaps (e.g., 429 retry handling, explicit bulk-output verification).

DimensionReasoningScore

Conciseness

The body is efficient and operational — wrapper CLI rules, rate limits, jq recipes, and context-management tactics Claude cannot know — with no explanations of concepts Claude already knows. It falls short of the 5 anchor due to trimmable redundancy: the "avoid one-by-one / prevents turn exhaustion and context overflow" guidance appears nearly verbatim in both Context Management and Bulk Workflows, the folder-structure listing duplicates the reference links in Functions, and the MANDATORY warning block is emphatic and padded. Score 3 would require more substantial unnecessary explanation, which is absent.

4 / 5

Actionability

Guidance is fully executable and copy-paste ready: `uv run scripts/pubmed_api.py ./search_results.json search_pubmed "BRCA1" --max_results 5`, the "Join PMIDs" chaining recipe, and null-safe jq slimming/filtering recipes cover the common search→fetch→process cases end-to-end. This matches the 5 anchor (specific examples cover the common cases); score 4 would imply gaps in the executable examples, which are not present.

5 / 5

Workflow Clarity

Multi-step workflows are clearly sequenced (search → join PMIDs → fetch → slim → read) with most checkpoints documented: non-zero exit-code error handling, "Filter Early" verification before reading into context, grounding rules, and explicit search-termination limits. It falls short of the 5 anchor because no full validate→fix→retry feedback loop exists (e.g., no retry guidance for 429 errors beyond obtaining an API key, and the bulk workflow lacks an explicit output-verification checkpoint); it stays above the 3 anchor because validation signals (exit codes, filter-early, termination limits) are explicitly stated rather than missing.

4 / 5

Progressive Disclosure

SKILL.md is a clear overview with one-line function tables, a mandatory read-before-call rule, and one-level-deep references that all exist and are well-signaled from the function section headers (verified against the actual bundle). Score 5 is blocked by minor navigation gaps: three reference files (advanced-linking.md, advanced-search.md, citation-matching.md) appear only in the flat folder-structure listing rather than being linked from their relevant sections — e.g., match_raw_citations points to utilities.md while the dedicated citation-matching.md guide is discoverable only via the folder listing. Score 3 would require content that should be separate inlined or buried references, which is not the case.

4 / 5

Total

17

/

20

Passed

Description

75%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 highly specific, comprehensive, third-person description of capabilities with excellent distinctiveness, held back only by the complete absence of explicit trigger guidance. Adding a "Use when..." clause would raise completeness substantially.

Suggestions

Add an explicit trigger clause, e.g.: "Use when the user asks to search PubMed, find biomedical papers or clinical trials, fetch abstracts or full text, link genes/compounds to published research, or resolve citations."

Broaden trigger vocabulary with natural synonyms users actually say — "papers", "biomedical literature", "medical research", "PMID" — to improve trigger_term_quality from good to comprehensive.

DimensionReasoningScore

Specificity

The description lists multiple concrete actions — "Search PubMed", "Fetch abstracts and full text", "Link published research to biological databases (gene, protein, nucleotide, PubChem)", "Verify medical spelling, match raw citations, and cache result sets" — comprehensively covering the skill's 10 CLI functions. This matches the top anchor (multiple specific concrete actions, comprehensive coverage); score 4 would require notable coverage gaps, which are absent.

5 / 5

Completeness

The "what" is clearly and comprehensively stated, but there is no "Use when..." clause or any equivalent explicit trigger guidance anywhere in the description, which caps completeness at 3 per the judging guidelines. Score 4 would require at least an implicit or weakly-specified "when", which is absent.

3 / 5

Trigger Term Quality

Good natural keyword coverage: "PubMed", "scientific literature", "clinical trials", "abstracts", "full text", "citations", "genes", "compounds", "PubChem" — phrases users would naturally say. It falls short of the 5 anchor because common synonyms and variants are missing (e.g., "papers", "biomedical literature", "PMID", "MEDLINE"), but is well above the 3 anchor's "missing common variations".

4 / 5

Distinctiveness Conflict Risk

The description occupies a clear niche with distinct triggers — "Search PubMed", "Interfaces NCBI E-utilities and PMC BioC APIs", "PubChem" — naming specific databases and APIs, so risk of triggering for a wrong (e.g., generic web-search or non-biomedical) skill is minimal. Score 4 would require minor overlap risk with closely related skills, which is not evident.

5 / 5

Total

17

/

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/pubmed_database/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.