sentence-transformers

Framework for state-of-the-art sentence, text, and image embeddings. Provides 5000+ pre-trained models for semantic similarity, clustering, and retrieval. Supports multilingual, domain-specific, and multimodal models. Use for generating embeddings for RAG, semantic search, or similarity tasks. Best for production embedding generation.

Quality

72%

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 ./backend/cli/skills/llm-tools/sentence-transformers/SKILL.md

Quality

Content

61%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.

Highly actionable with broad, executable code examples, but it is held back by promotional padding, triplicated model-selection content, a complete lack of validation checkpoints in the fine-tuning workflow, and an orphaned references/models.md file that is never linked from the body.

Suggestions

Link and defer to references/models.md for the model catalog: replace the inlined 'Popular models', 'Model selection guide', and 'Performance' tables with a short pointer (e.g., 'See [models.md](references/models.md) for the full model selection matrix and benchmarks') to eliminate duplication and fix the orphaned reference.

Remove the promotional 'Metrics' block (GitHub stars, language counts, 'Based on PyTorch/Transformers') — it does not help Claude act and wastes context tokens.

Add an explicit validation/feedback loop to the fine-tuning workflow (e.g., evaluate on a held-out set after model.fit, retrain if quality drops) so the multi-step process meets the workflow-clarity bar above 3.

Dimension	Reasoning	Score
Conciseness	Mostly concrete code, but it carries padding that does not earn its place — a promotional 'Metrics' block ('15,700+ GitHub stars', 'Based on PyTorch/Transformers') and model-selection content duplicated across 'Popular models', the 'Model selection guide' table, and the 'Performance' table, fitting the 'mostly efficient but includes some unnecessary explanation or could be tightened' anchor rather than the lean 4.	3 / 5
Actionability	Provides fully executable, copy-paste-ready code across the common cases — install, basic encoding, cosine similarity, semantic search, batch encoding, fine-tuning, and LangChain/LlamaIndex integration — matching the 'fully executable; copy-paste ready; covers common cases' anchor.	5 / 5
Workflow Clarity	The body is a collection of independent recipes rather than a sequenced process; the fine-tuning section implies a multi-step flow (data → dataloader → loss → fit → save) but has no explicit validation checkpoint or feedback loop, and batch operations lack verification, so per the feedback-loop cap it cannot exceed 3.	3 / 5
Progressive Disclosure	A bundle file references/models.md exists but is never linked or referenced from the body, while the model-guide content it should hold is inlined and duplicated in the SKILL.md tables — the classic 'content that clearly belongs in separate files is inlined; references are buried' pattern, which is worse than merely under-signaled.	2 / 5
	Total	13 / 20 Passed

Description

83%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, third-person description that explicitly covers both what the skill does and when to use it with concrete trigger phrases. Minor gaps in capability specificity and a few missing trigger synonyms prevent a perfect score.

Dimension	Reasoning	Score
Specificity	Names the domain ('sentence, text, and image embeddings') and lists several concrete capabilities — semantic similarity, clustering, retrieval, RAG, semantic search — but stops short of fully comprehensive coverage of the framework's operations (e.g., fine-tuning is omitted), fitting the 'several specific actions; minor gaps' anchor rather than the comprehensive 5.	4 / 5
Completeness	It clearly answers 'what' ('Framework for... embeddings', 'Provides 5000+ pre-trained models', 'Supports multilingual, domain-specific, and multimodal models') and explicitly answers 'when' with concrete trigger phrases ('Use for generating embeddings for RAG, semantic search, or similarity tasks'), matching the top anchor; the 'Use for...' clause satisfies the explicit-trigger requirement so completeness is not capped.	5 / 5
Trigger Term Quality	Includes natural terms users actually say ('embeddings', 'RAG', 'semantic search', 'semantic similarity', 'clustering', 'multilingual', 'multimodal') with good coverage, but a few common variants (e.g., 'vector search', 'text embeddings', 'similarity scores') are absent, matching the 'good keyword coverage; a few natural terms missing' anchor.	4 / 5
Distinctiveness Conflict Risk	It carves a clear local-embedding niche and even names alternatives (OpenAI Embeddings, Instructor, Cohere Embed) to reduce overlap, but because several sibling embedding skills plausibly exist, residual minor overlap risk keeps it at 'mostly distinct' rather than the minimal-conflict 5.	4 / 5
	Total	17 / 20 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/llm-tools/sentence-transformers/SKILL.md
Commit: 1d182e9

Reviewed: 3 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.