nemo-guardrails

NVIDIA's runtime safety framework for LLM applications. Features jailbreak detection, input/output validation, fact-checking, hallucination detection, PII filtering, toxicity detection. Uses Colang 2.0 DSL for programmable rails. Production-ready, runs on T4 GPU.

Quality

66%

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/llm-tools/nemo-guardrails/SKILL.md

Quality

Content

57%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 code-rich and largely actionable, covering the main safety workflows with concrete examples. It is held back by time-sensitive padding, example-style rather than checkpointed workflows, and three referenced reference files that are missing from the bundle.

Suggestions

Remove or relocate time-sensitive details (version strings, star counts, latency tables) into a versioned/deprecated section or drop them, to improve conciseness.

Provide the missing referenced files (references/colang-guide.md, references/integrations.md, references/performance.md) or remove the broken links.

Reframe the workflows as explicit sequenced steps with validation checkpoints (e.g., verify config loads, test refusal on a known jailbreak input) rather than standalone code examples.

Dimension	Reasoning	Score
Conciseness	Mostly efficient code-driven content, but it includes time-sensitive filler ('Version: v0.9.0+ (v0.12.0 expected)', 'GitHub ⭐ 4,300+') and verbose hardware/latency tables that pad the body without adding guidance Claude needs.	3 / 5
Actionability	Provides numerous concrete, mostly copy-paste-ready code blocks across the common cases, with only minor gaps where helper calls like toxicity_detector() and verify_facts() are used without definition.	4 / 5
Workflow Clarity	The five 'workflows' are presented as illustrative code examples rather than sequenced processes with explicit validation checkpoints or feedback loops, so the sequence is present but checkpoints are implicit.	3 / 5
Progressive Disclosure	Structure is reasonable and references to colang-guide.md, integrations.md, and performance.md are clearly signaled one level deep, but those referenced files do not exist in the bundle, so navigation fails in practice.	3 / 5
	Total	13 / 20 Passed

Description

75%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 distinctive, naming a concrete framework and a comprehensive set of safety capabilities in third person. Its main weakness is the absence of an explicit 'Use when...' trigger clause, which caps completeness.

Suggestions

Add an explicit 'Use when...' clause naming natural trigger phrases (e.g., 'Use when adding runtime safety checks, jailbreak detection, or PII filtering to LLM applications').

Include a few natural synonyms users might say ('prompt injection', 'content moderation', 'safety rails') to broaden trigger coverage.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete capabilities — 'jailbreak detection, input/output validation, fact-checking, hallucination detection, PII filtering, toxicity detection' — plus the Colang 2.0 DSL mechanism, giving comprehensive coverage of what the framework does.	5 / 5
Completeness	The 'what' is clearly and concretely stated, but there is no 'Use when...' clause or equivalent explicit trigger guidance, which caps completeness at 3 per the rubric.	3 / 5
Trigger Term Quality	Natural terms users would say are well covered ('jailbreak detection', 'PII filtering', 'hallucination detection', 'toxicity detection', 'guardrails'), but a few common synonyms or phrasings users might naturally invoke are absent.	4 / 5
Distinctiveness Conflict Risk	Names a specific NVIDIA framework (NeMo Guardrails) with a distinct runtime-safety niche and Colang DSL triggers, giving minimal overlap with other skills.	5 / 5
	Total	17 / 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
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 3 missing	Warning
referenced_paths_exist	Referenced path issues: 6 missing	Warning

	Total	13 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/llm-tools/nemo-guardrails/SKILL.md
Commit: fcd2e11

Reviewed: 2 days 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.