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

70%

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

65%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 content is highly actionable with abundant executable code, but it duplicates material that belongs in references/models.md and never links to that file. Workflow clarity is middling because batch and fine-tuning operations lack validation checkpoints or error-recovery loops.

Suggestions

Link the 'Popular models', 'Model selection guide', and 'Performance' sections to references/models.md (e.g., 'See [models.md](references/models.md) for the full model catalog') instead of inlining duplicate tables.

Add a validation/verification checkpoint to the fine-tuning and batch-encoding workflows (e.g., check loss is decreasing, verify embedding shapes before saving).

Trim the repeated model/dimension/speed information between the inline tables and models.md to reduce token overhead.

Dimension	Reasoning	Score
Conciseness	Mostly efficient with runnable code, but sections like 'When to use', 'Popular models', 'Model selection guide', and 'Performance' repeat overlapping model/speed/dimension information already covered in references/models.md, adding padding.	3 / 5
Actionability	Fully executable, copy-paste-ready code across installation, basic usage, semantic search, batch encoding, fine-tuning, and LangChain/LlamaIndex integration, covering the common cases.	5 / 5
Workflow Clarity	The skill involves batch encoding and fine-tuning but presents steps without validation checkpoints or error-recovery feedback loops; the 'Best practices' list is advisory rather than a sequenced workflow.	3 / 5
Progressive Disclosure	A references/models.md bundle exists and is one level deep, but the body never signals or links to it—model and performance tables are inlined instead of pointing to the reference, so references are present but not clearly signaled.	3 / 5
	Total	14 / 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.

The description is specific and complete, clearly stating both what the framework does and when to use it with concrete trigger terms like RAG and semantic search. It is slightly held back by some missing natural synonyms and minor overlap risk with adjacent embedding/vector skills.

Suggestions

Add a more explicit 'Use when the user asks for...' trigger clause naming concrete situations (e.g., 'Use when the user needs to embed documents for a vector database or compare text similarity').

Include natural synonyms users might say such as 'vector search' or 'text similarity' alongside 'semantic search'.

Sharpen distinctiveness by noting the local/no-API angle in the description itself rather than only the body.

Dimension	Reasoning	Score
Specificity	Lists several concrete actions—'generating embeddings for RAG, semantic search, or similarity tasks' plus 'semantic similarity, clustering, and retrieval'—with only minor coverage gaps (no explicit mention of fine-tuning or cross-encoder use cases).	4 / 5
Completeness	It clearly answers 'what' ('Framework for...embeddings...5000+ pre-trained models') and 'when' ('Use for generating embeddings for RAG, semantic search, or similarity tasks'), though the 'when' clause could be more explicit about the specific user situations that call for it.	4 / 5
Trigger Term Quality	Natural trigger terms are present ('RAG', 'semantic search', 'similarity tasks', 'semantic similarity', 'embeddings') but a few common phrasings users might say ('vector search', 'text similarity') are missing.	4 / 5
Distinctiveness Conflict Risk	The embedding/RAG niche is mostly distinct with clear triggers, but it could overlap with a general 'HuggingFace models' or 'vector search' skill, leaving minor conflict risk.	4 / 5
	Total	16 / 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: 3a6c3a9

Reviewed: 1 day 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.