clip

OpenAI's model connecting vision and language. Enables zero-shot image classification, image-text matching, and cross-modal retrieval. Trained on 400M image-text pairs. Use for image search, content moderation, or vision-language tasks without fine-tuning. Best for general-purpose image understanding.

Quality

78%

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

Fix and improve this skill with Tessl

tessl review fix ./backend/cli/skills/llm-tools/clip/SKILL.md

Quality

Content

68%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 actionable, code-rich skill body with strong examples, undermined by duplicated application content that belongs in the unused references/applications.md bundle and by missing validation in its batch workflows.

Suggestions

Move the application recipes (zero-shot classification, semantic search, content moderation) into references/applications.md and link to it from a short "Applications" section instead of inlining them, removing the duplication.

Add a validation/verification checkpoint to the batch-processing and vector-DB indexing examples (e.g., assert shapes, check embedding norms, or verify retrieved results) to satisfy the batch-operation feedback-loop expectation.

Trim the redundant one-line intro and the promotional "Metrics" block (GitHub stars) and present the model lineup once — either the Python list or the table — to tighten conciseness.

Dimension	Reasoning	Score
Conciseness	Mostly lean code recipes that assume Claude's competence, but models are shown twice (Python list plus table), the "Metrics" block carries promotional GitHub-star fluff, and a redundant one-line intro repeats the description.	4 / 5
Actionability	Numerous complete, copy-paste-ready Python examples cover the common cases — zero-shot classification, image-text similarity, semantic search, batch processing, and vector-DB integration — with executable detail beyond minor gaps.	5 / 5
Workflow Clarity	The skill is a recipe collection rather than a sequenced process, and batch operations (batch processing, vector-DB indexing) lack any validation or verification checkpoints, which caps workflow clarity per the batch-operation guideline.	3 / 5
Progressive Disclosure	A bundle file references/applications.md exists but is never linked from the body, and the body inlines application recipes (zero-shot classification, semantic search, content moderation) that duplicate that file — content that clearly belongs in the reference is inlined and the reference is not signaled.	2 / 5
	Total	14 / 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 that states concrete capabilities and an explicit "Use for" trigger clause with specific use cases. It would reach full marks by adding more user-natural synonyms and tightening the over-broad closing line.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — "zero-shot image classification, image-text matching, and cross-modal retrieval" plus "image search, content moderation" — with comprehensive coverage rather than minor gaps.	5 / 5
Completeness	Explicitly answers both what ("model connecting vision and language" with the enabled actions) and when ("Use for image search, content moderation, or vision-language tasks without fine-tuning") with concrete triggers.	5 / 5
Trigger Term Quality	Natural terms like "image search", "content moderation", and "image classification" are present, but it leans on jargon ("cross-modal retrieval", "vision-language tasks") and lacks synonym/extension coverage, stopping short of comprehensive.	4 / 5
Distinctiveness Conflict Risk	The zero-shot-without-fine-tuning CLIP niche is distinct, but the broad "general-purpose image understanding" and shared triggers like "image search"/"content moderation" create minor overlap risk with BLIP-2/LLaVA.	4 / 5
	Total	18 / 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/clip/SKILL.md
Commit: fcd2e11

Reviewed: 2 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.