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

75%

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 ./18-multimodal/clip/SKILL.md

Quality

Content

65%

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 strong executable examples, but it is padded with non-evergreen metrics and leaves its reference bundle unlinked. Batch/moderation workflows also lack validation checkpoints.

Suggestions

Link the existing references/applications.md from the body (e.g. under a '## Applications' or '## Advanced' section) so the bundle is discoverable and signaling is explicit.

Remove or relocate time-sensitive, non-evergreen content such as GitHub star counts and the 'Metrics' block; if kept, place version/date-sensitive details in a dedicated deprecated/notes section per the rubric.

Add validation/verification checkpoints to batch and content-moderation workflows (e.g. confirming probabilities are sensible, sanity-checking top-k results) to support feedback loops for batch operations.

Dimension	Reasoning	Score
Conciseness	The body is mostly lean and code-driven, but contains padding Claude doesn't need ("OpenAI's model that understands images from natural language", a "Metrics" block with "25,300+ GitHub stars") and time-sensitive claims (star counts, version-style stats) without a deprecated/old-patterns section, fitting the 'mostly efficient but includes some unnecessary explanation or could be tightened' anchor rather than the lean level 3.	2 / 3
Actionability	Multiple fully executable, copy-paste-ready code blocks cover zero-shot classification, similarity, search, moderation, batching, and vector-DB integration, matching the level-3 anchor of fully executable, specific, copy-paste-ready examples.	3 / 3
Workflow Clarity	Sections are clearly sequenced by use case, but batch and moderation operations have no validation/verification checkpoints or feedback loops; per the rubric, missing validation in batch operations caps workflow clarity at 2 rather than allowing 3.	2 / 3
Progressive Disclosure	The body has well-organized sections but a provided bundle file (references/applications.md) is never linked or signaled anywhere in the body, and substantial material (model tables, performance numbers) stays inline that could be split out — matching the 'some structure but references not clearly signaled / content that should be separate is inline' anchor, not the well-signaled one-level-deep level 3.	2 / 3
	Total	9 / 12 Passed

Description

85%

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, complete, and distinct, with a clear 'Use for...' trigger clause. Its weakest point is trigger-term coverage, which is good but misses several natural phrasings a user might say.

Suggestions

Broaden trigger terms to include phrasings users naturally say, e.g. "image-text similarity", "image retrieval", or "image-to-text matching", alongside the current ones.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions: "zero-shot image classification", "image-text matching", "cross-modal retrieval", "image search", "content moderation" — matching the level-3 anchor of multiple specific concrete actions, not level 2 which only names a domain and some actions.	3 / 3
Completeness	Explicitly answers what it does ("connecting vision and language", the listed capabilities) and when to use it ("Use for image search, content moderation, or vision-language tasks without fine-tuning"), matching the level-3 anchor that clearly answers both what AND when with explicit triggers.	3 / 3
Trigger Term Quality	Natural terms like "image search", "content moderation", "image classification" are present, but common variations users might say (e.g. "image-text similarity", "image retrieval", "vision-language") are only partially covered, fitting the 'some relevant keywords but missing common variations' anchor rather than the full-coverage level 3.	2 / 3
Distinctiveness Conflict Risk	The "Trained on 400M image-text pairs" framing and CLIP-specific multimodal niche give it a clear, distinct trigger space unlikely to conflict with other skills, matching the level-3 clear-niche anchor.	3 / 3
	Total	11 / 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: Orchestra-Research/AI-Research-SKILLs
Path: 18-multimodal/clip/SKILL.md
Commit: 773a529

Reviewed: about 12 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.