llava

Large Language and Vision Assistant. Enables visual instruction tuning and image-based conversations. Combines CLIP vision encoder with Vicuna/LLaMA language models. Supports multi-turn image chat, visual question answering, and instruction following. Use for vision-language chatbots or image understanding tasks. Best for conversational image analysis.

Quality

67%

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/llava/SKILL.md

Quality

Content

50%

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

The body is well-organized with strong executable Quick start and deployment examples, but it is held back by redundant common-task snippets, pseudocode helpers in the conversational sections, and a training reference that exists as a bundle file yet is never actually referenced from the SKILL.md.

Suggestions

Replace the undefined ask()/generate() helpers in the multi-turn and common-tasks sections with the actual executable inference code used in Quick start, or consolidate them into one concrete example.

Collapse the five near-identical "Common tasks" snippets into a single example plus a concise list of question phrasings to remove token-inefficient padding.

Link the existing references/training.md from the "Training custom model" section (e.g. "See [training.md](references/training.md) for full stages, data format, and LoRA") so the bundle file is clearly signaled and one level deep.

Dimension	Reasoning	Score
Conciseness	Mostly efficient with executable code and tables, but the five near-identical "Common tasks" snippets are token-inefficient padding and several sections use undefined helper functions (ask()/generate()) as pseudocode instead of lean executable code.	2 / 3
Actionability	The Quick start, CLI, Gradio, and quantization sections give fully executable code, but the multi-turn and common-tasks sections rely on undefined ask()/generate() helpers rather than complete copy-paste-ready code, leaving guidance incomplete.	2 / 3
Workflow Clarity	Installation-to-inference and the two training stages are sequenced, but there are no explicit validation or verification checkpoints (e.g. confirming a model loads or a response is valid) for the batch/destructive-adjacent operations described.	2 / 3
Progressive Disclosure	The body is sectioned, but the provided bundle file references/training.md is never linked or signaled from the body — the "Training custom model" section inlines bash commands that overlap the reference instead of pointing to it, so navigation to the deeper material is missing.	2 / 3
	Total	8 / 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 and complete, clearly stating both capabilities and an explicit use-trigger for a well-defined vision-language niche. Its main weakness is trigger-term breadth, omitting several natural phrasings users would actually say.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions including "visual instruction tuning", "image-based conversations", "multi-turn image chat", "visual question answering", and "instruction following" rather than vague language.	3 / 3
Completeness	Explicitly answers both what it does (capabilities listed) and when to use it via the explicit "Use for vision-language chatbots or image understanding tasks" trigger clause.	3 / 3
Trigger Term Quality	Includes relevant terms like "visual question answering" and "image understanding", but misses common natural variations a user would say (e.g. "VQA", "image chat", "describe an image", "image captioning").	2 / 3
Distinctiveness Conflict Risk	Occupies a clear vision-language niche with distinct triggers, making it unlikely to fire for unrelated skills.	3 / 3
	Total	11 / 12 Passed

Validation

75%

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 — 12 / 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
referenced_paths_exist	Referenced path issues: 2 missing, 2 deeper-than-1-level	Warning

	Total	12 / 16 Passed

Repository: NousResearch/hermes-agent
Path: optional-skills/mlops/llava/SKILL.md
Commit: 4c9628e

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.