Hub Validators
Browse and install validators from the Guardrails Hub
✅ Hub Token configured. You have access to all validators. Change token →
How to install: Click ⬇ Install on any validator. It will be installed into the Guardrails API container. After installing, click 🔄 Restart API to make it available, then add it to a Guard using its ID.

Content Safety

Toxic Language
hub://guardrails/toxic_language
Detects toxic, offensive, or harmful language in text output.
Profanity Free
hub://guardrails/profanity_free
Ensures text is free from profanity and vulgar language.
NSFW Text
hub://guardrails/nsfw_text
Detects not-safe-for-work content in text responses.
Gibberish
hub://guardrails/gibberish
Detects nonsensical or gibberish text in model outputs.
Unusual Prompt
hub://guardrails/unusual_prompt
Flags prompts that are unusually structured or potentially adversarial.
Bias Check
hub://guardrails/bias_check
Checks for various forms of bias (gender, race, political) in text.
Hate Speech
hub://guardrails/hate_speech
Identifies hate speech targeting individuals or groups.

Security

Prompt Injection
hub://guardrails/detect_prompt_injection
Detects prompt injection attempts in user inputs.
SQL Injection
hub://guardrails/sql_injection
Detects potential SQL injection strings in text.
Secrets Present
hub://guardrails/secrets_present
Detects API keys, passwords, tokens, or other secrets in output.
LLM Critique
hub://guardrails/llm_critique
Uses a second LLM call to critique and validate the first response.

Privacy

Detect PII
hub://guardrails/detect_pii
Detects personally identifiable information (names, emails, SSNs, etc).
Anonymize PII
hub://guardrails/anonymize_pii
Detects and anonymizes PII by replacing with placeholders.
Sensitive Topics
hub://guardrails/sensitive_topics
Flags mentions of pre-configured sensitive topics.

Format

Valid JSON
hub://guardrails/valid_json
Ensures the output is valid, parseable JSON.
Valid Python
hub://guardrails/valid_python
Ensures the output is syntactically valid Python code.
Valid SQL
hub://guardrails/valid_sql
Ensures the output is a valid SQL statement.
Regex Match
hub://guardrails/regex_match
Validates that text matches a user-specified regular expression.
Valid Length
hub://guardrails/valid_length
Ensures text stays within minimum and maximum character limits.
Ends With
hub://guardrails/ends_with
Validates that text ends with a specific string.
One Line
hub://guardrails/one_line
Ensures the response is a single line with no newlines.
Valid Choices
hub://guardrails/valid_choices
Ensures output is one of a pre-defined set of allowed values.
Valid Range
hub://guardrails/valid_range
Ensures numeric output falls within a specified min/max range.
Valid OpenAPI
hub://guardrails/valid_openapi
Validates that the output is a valid OpenAPI specification.

Quality

QA Relevance
hub://guardrails/qa_relevance_without_reference_answer
Validates that an answer is relevant to the question asked.
Sycophancy Check
hub://guardrails/sycophancy_check
Detects sycophantic or excessively agreeable AI responses.
On Topic
hub://guardrails/on_topic
Ensures the response stays on the expected topic or domain.
Reading Time
hub://guardrails/reading_time
Validates that text falls within an expected reading time window.
Similar to Document
hub://guardrails/similar_to_document
Checks that response is semantically similar to a reference document.
Response Evaluator
hub://guardrails/response_evaluator
Uses an LLM to evaluate response quality against custom criteria.

Business

Competitor Check
hub://guardrails/competitor_check
Flags if a competitor brand is mentioned in the response.
Restrict to Topic
hub://guardrails/restrict_to_topic
Limits responses to a specific business domain or topic area.
Translation Quality
hub://guardrails/translation_quality
Validates the quality of translated text using back-translation.
URL Reachability
hub://guardrails/url_reachability
Checks that URLs in the response are valid and reachable.