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

82%

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

65%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A code-rich, highly actionable skill body held back by padding, absent validation checkpoints, and an orphaned reference file. The inline content is strong but redundant with the unused applications.md bundle.

Suggestions

Replace the marketing "Metrics" block and trim the Resources/Limitations tail to only what Claude needs to apply the skill, removing content that does not earn its tokens.

Add explicit validation/verification steps to the batch-processing and vector-database workflows (e.g., confirm embedding shapes, verify collection counts after insert, retry on failure) to lift workflow clarity.

Link references/applications.md from the body (e.g., a "## Advanced applications → See applications.md" section) and move the duplicated use-case recipes out of SKILL.md so the main file is a lean overview with one-level-deep references.

Dimension	Reasoning	Score
Conciseness	The body is mostly efficient executable code, but the marketing "Metrics" block ("25,300+ GitHub stars", "Trained on 400M image-text pairs") and a Best practices/Limitations/Resources tail pad it unnecessarily, matching the mostly-efficient-but-could-be-tightened score-2 anchor rather than the lean score-3 anchor.	2 / 3
Actionability	It provides abundant copy-paste-ready Python (zero-shot classification, similarity, search, moderation, batch, vector-DB integration) plus install commands, matching the fully-executable score-3 anchor.	3 / 3
Workflow Clarity	Each recipe has a clear load→preprocess→encode→similarity sequence, but batch processing and vector-DB insertion lack any validation/verification checkpoints, so per the destructive/batch cap workflow clarity stays at the score-2 anchor rather than reaching score-3.	2 / 3
Progressive Disclosure	A references/applications.md bundle exists but is never linked or signaled from the body, and the same use cases (zero-shot, semantic search, content moderation, best practices) are duplicated inline, matching the references-present-but-not-signaled score-2 anchor rather than the well-signaled one-level-deep score-3 anchor.	2 / 3
	Total	9 / 12 Passed

Description

100%

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 names concrete capabilities and gives explicit "Use for" trigger guidance covering both what and when. Minor jargon (cross-modal retrieval) aside, it is specific, complete, and distinct.

Dimension	Reasoning	Score
Specificity	"Enables zero-shot image classification, image-text matching, and cross-modal retrieval" plus "image search, content moderation" lists multiple concrete actions, matching the score-3 anchor rather than the domain-but-incomplete score-2 anchor.	3 / 3
Completeness	It states what the model does and includes an explicit "Use for image search, content moderation, or vision-language tasks" trigger clause, satisfying both what AND when per the score-3 anchor rather than the when-implied score-2 anchor.	3 / 3
Trigger Term Quality	Natural terms a user would say (image search, image classification, content moderation, image-text matching) are well covered; while "cross-modal retrieval" is jargon, the overall coverage matches the score-3 anchor and exceeds the partial score-2 anchor.	3 / 3
Distinctiveness Conflict Risk	CLIP-specific triggers (zero-shot image classification, cross-modal retrieval, image-text matching) carve a clear niche unlikely to conflict with other skills, matching the score-3 anchor; "general-purpose image understanding" adds slight breadth but not enough to drop to score-2.	3 / 3
	Total	12 / 12 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: 52845c3

Reviewed: about 13 hours 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.