clip

Zero-shot image classification and image-text search.

Quality

63%

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 ./optional-skills/mlops/clip/SKILL.md

Quality

Content

72%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 executable code for the common CLIP use cases and stays mostly lean, but it functions as a recipe grab-bag without validation checkpoints and fails to surface its bundled applications reference. Linking the reference file and adding light validation steps would lift the weaker dimensions.

Suggestions

Reference references/applications.md explicitly from the body (e.g., a '## More applications' section linking to it) and de-duplicate the overlapping classification/search/moderation recipes.

Add a validation checkpoint to the batch and content-moderation flows (e.g., a confidence threshold with a fallback/uncertain branch) to lift workflow clarity above 3.

Trim obvious inline comments like '# Load model' and '# Cosine similarity' that restate what the code already shows.

Dimension	Reasoning	Score
Conciseness	The body is code-forward and largely avoids explaining concepts Claude already knows, with only minor over-explanation in inline comments (e.g., '# Load model', '# Cosine similarity') that could be trimmed; it does not reach 5 because of those small padded spots and some redundancy with the bundled applications reference.	4 / 5
Actionability	It provides copy-paste-ready, fully executable code across the common cases (zero-shot classification, similarity, search, moderation, batch, vector-DB integration) plus a model table, matching the fully-executable anchor.	5 / 5
Workflow Clarity	Each recipe is internally clear, but the skill is a collection of independent snippets rather than a sequenced workflow, and the batch-processing section lacks any validation/verification checkpoint, which per the rubric caps workflow clarity at 3.	3 / 5
Progressive Disclosure	Section headers give reasonable structure, but the bundled references/applications.md is never linked from the body and duplicates inlined content (classification, search, moderation), fitting the 'references present but not clearly signaled; content that should be separate is inline' anchor.	3 / 5
	Total	15 / 20 Passed

Description

53%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 concise and names two concrete capabilities but omits any explicit trigger guidance, capping completeness and trigger-term quality at the mid level. It is reasonably distinct yet would benefit from a 'Use when...' clause and more natural synonyms.

Suggestions

Add an explicit 'Use when...' clause naming natural triggers (e.g., 'Use when classifying images without training data or searching images with text queries').

Include synonyms and concrete phrasings users actually say, such as 'image search', 'text-to-image matching', and 'classify images without labels'.

Broaden the action list slightly (e.g., add image-text similarity scoring) to push specificity toward comprehensive coverage.

Dimension	Reasoning	Score
Specificity	The phrase 'Zero-shot image classification and image-text search' names the domain plus two concrete actions, matching the anchor for 1-2 named actions without comprehensive coverage; it does not reach 4 because only two actions are listed.	3 / 5
Completeness	It gives a clear 'what' (zero-shot image classification, image-text search) but no 'Use when...' or equivalent trigger guidance, so per the cap it stays at 3 rather than 4.	3 / 5
Trigger Term Quality	Terms like 'image classification' and 'image-text search' are relevant, but natural user phrasings (e.g., 'classify images without training', 'search images with text') and synonyms/extensions are missing, fitting the 'some relevant keywords but missing variations' anchor.	3 / 5
Distinctiveness Conflict Risk	'Zero-shot image classification' signals a fairly distinct CLIP-style niche with only minor overlap against closely related vision-language models, landing mostly-distinct rather than fully-distinct because image-text search overlaps with general retrieval skills.	4 / 5
	Total	13 / 20 Passed

Validation

81%

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 — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	13 / 16 Passed

Repository: NousResearch/hermes-agent
Path: optional-skills/mlops/clip/SKILL.md
Commit: 6680afb

Reviewed: about 3 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.