llama-factory

Expert guidance for fine-tuning LLMs with LLaMA-Factory - WebUI no-code, 100+ models, 2/3/4/5/6/8-bit QLoRA, multimodal support

Quality

49%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./backend/cli/skills/ml-training/llama-factory/SKILL.md

Quality

Content

28%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 body is a scraped-documentation template shell with no domain content: no commands, no examples, and no guidance beyond pointing at generically described reference files. Structure is superficially sound (overview plus one-level references), but placeholder sections, an omitted reference file, and pointers to nonexistent files undermine both actionability and navigation.

Suggestions

Replace the placeholder Quick Reference section with 2-3 real, executable commands (e.g. `llamafactory-cli webui`, a sample training yaml and `llamafactory-cli train`) instead of 'patterns will be added'.

Give each reference file a substantive description of its actual contents and list all 5 files including index.md; remove references to nonexistent 'tutorials', 'api', and 'guides' files.

Delete the boilerplate Notes/Resources/Updating sections ('Add helper scripts here', 'Re-run the scraper') that contribute no actionable content, or replace them with skill-specific maintenance guidance.

Dimension	Reasoning	Score
Conciseness	Nearly every section is template boilerplate or placeholder ('Quick reference patterns will be added as you use the skill', 'Add helper scripts here', 'Add templates, boilerplate, or example projects here') that adds no knowledge. It is padded rather than explaining known concepts, so it sits above the severely-verbose anchor but well below efficient.	2 / 5
Actionability	Minimal concrete guidance: the only instructions are pointers to real reference files ('Start with the getting_started... reference files'), plus a nonexistent command ('Use `view` to read specific reference files') and the vague 're-run the scraper'. No code, commands, or executable steps appear anywhere in the body.	2 / 5
Workflow Clarity	A rough sequence exists (trigger conditions, then beginners start with getting_started, then category references), but it references nonexistent files ('tutorials', 'api', 'guides') and includes no validation or checkpoints of any kind.	2 / 5
Progressive Disclosure	The split itself is appropriate: SKILL.md is an overview and bulk content lives in one-level-deep reference files. But the bundle has 5 reference files and the body lists only 4 (omitting index.md), and the one-line descriptions ('Advanced documentation', 'Other documentation') carry no navigational value, so references are present but not clearly signaled.	3 / 5
	Total	9 / 20 Passed

Description

70%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 specific and distinct, naming concrete LLaMA-Factory capabilities and natural trigger keywords. Its main weakness is the complete absence of explicit 'when to use' guidance, which caps completeness and weakens its value as a trigger description.

Suggestions

Append a 'Use when...' clause, e.g. 'Use when fine-tuning or QLoRA-tuning LLMs with LLaMA-Factory, launching its WebUI, or training multimodal models.'

Convert noun-phrase capability lists into concrete actions (e.g. 'fine-tune', 'export and merge adapters', 'evaluate models') for a stronger 'what' statement.

Add common synonyms such as 'train', 'adapter', 'PEFT', and 'instruct tuning' to broaden natural trigger-term coverage.

Dimension	Reasoning	Score
Specificity	Lists several concrete capabilities ('WebUI no-code', '2/3/4/5/6/8-bit QLoRA', 'multimodal support', '100+ models') but presents them as noun phrases rather than actions and omits what the skill actually does (train, configure, export, merge adapters).	4 / 5
Completeness	The 'what' is clear (fine-tuning LLMs with LLaMA-Factory and its feature set), but there is no 'Use when...' clause or any trigger guidance, which caps completeness at 3 per the rubric guidelines.	3 / 5
Trigger Term Quality	Good coverage of natural terms users would say ('fine-tuning', 'LLaMA-Factory', 'QLoRA', 'LoRA', 'WebUI', 'multimodal', 'HuggingFace'), but missing common variations like 'train a model', 'adapter', 'PEFT', or 'instruct tuning'.	4 / 5
Distinctiveness Conflict Risk	Names a specific tool with niche-specific details (bit-width QLoRA, WebUI no-code), giving it a clear niche with distinct triggers and minimal conflict risk against generic ML or fine-tuning skills.	5 / 5
	Total	16 / 20 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
metadata_version	'metadata.version' is missing	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/ml-training/llama-factory/SKILL.md
Commit: 4082a2e

Reviewed: 1 day 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.