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

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 ./backend/cli/skills/llm-tools/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 leads with solid executable quick-start code but is undermined by padding (repeated task snippets, non-functional integration stubs), undefined helper functions in several examples, no validation steps for training/batch workflows, and an unreferenced training bundle file.

Suggestions

Replace the five identical 'Common tasks' snippets and the LangChain `# Custom LLaVA inference` stub with one concrete, runnable example or remove them.

Define (or inline) the `ask`, `generate`, and `ask_llava` helpers used in the multi-turn and integration sections so the code is copy-paste ready.

Link the body's 'Training custom model' section to `references/training.md` and add a validation/checkpoint step to the training and batch-processing workflows.

Dimension	Reasoning	Score
Conciseness	The quick-start and CLI sections are lean, but the 'Common tasks' block repeats five near-identical `ask(model, image, question)` snippets and the LangChain/Gradio integration stubs add filler, so it is mostly efficient but could be tightened.	2 / 3
Actionability	The quick-start provides fully executable code, but 'Common tasks', multi-turn, and LangChain sections rely on undefined helpers (`ask`, `generate`, `ask_llava`) and a `# Custom LLaVA inference` stub, leaving key examples incomplete rather than copy-paste ready.	2 / 3
Workflow Clarity	Steps are sequenced (multi-turn turns, two training stages), but there are no validation/verification checkpoints for risky batch or training operations, which caps workflow clarity at 2 per the rubric.	2 / 3
Progressive Disclosure	A bundle file `references/training.md` exists with detailed training content, but the body's 'Training custom model' section never links to it, and the body itself is a long monolithic sweep of many topics that could be split out.	2 / 3
	Total	8 / 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 states concrete capabilities and provides an explicit 'Use for' trigger clause covering natural user phrasings. It is clearly scoped to vision-language tasks with low conflict risk.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions: 'Enables visual instruction tuning and image-based conversations', 'Supports multi-turn image chat, visual question answering, and instruction following' — matching the 'multiple specific concrete actions' anchor.	3 / 3
Completeness	Answers both what ('Enables... Supports...') and when with an explicit trigger clause ('Use for vision-language chatbots or image understanding tasks. Best for conversational image analysis.'), satisfying the explicit-trigger requirement.	3 / 3
Trigger Term Quality	Covers natural terms users would say — 'image-based conversations', 'visual question answering', 'multi-turn image chat', 'image understanding tasks', 'conversational image analysis' — giving good coverage of common phrasings.	3 / 3
Distinctiveness Conflict Risk	Occupies a clear multimodal vision-language niche with distinct triggers (image chat, VQA, image understanding), making it unlikely to fire for non-vision skills.	3 / 3
	Total	12 / 12 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
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	14 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/llm-tools/llava/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.