BerriAI/litellm · error · ValueError

Unknown pattern name: '{pattern_name}'. Available patterns:

Error message

Unknown pattern name: '{pattern_name}'. Available patterns: {available_patterns}

What it means

get_compiled_pattern() resolves a masking rule's pattern by name against PREBUILT_PATTERNS, which is loaded from the patterns.json shipped with the guardrail (us_ssn, email, credit_card, github_token, generic_api_key, iban, sg_nric, ...). A name that is not a key in that file raises ValueError listing every valid name, so the fix is always visible in the message itself.

Source

Thrown at litellm/proxy/guardrails/guardrail_hooks/litellm_content_filter/patterns.py:73

    PATTERN_EXTRA_CONFIG[pattern_data["name"]] = extra_config


def get_compiled_pattern(pattern_name: str) -> Pattern[str]:
    """
    Get a compiled regex pattern by name.

    Args:
        pattern_name: Name of the prebuilt pattern

    Returns:
        Compiled regex pattern

    Raises:
        ValueError: If pattern_name is not found in PREBUILT_PATTERNS
    """
    if pattern_name not in PREBUILT_PATTERNS:
        available_patterns: Final = ", ".join(PREBUILT_PATTERNS.keys())
        raise ValueError(f"Unknown pattern name: '{pattern_name}'. Available patterns: {available_patterns}")

    return re.compile(PREBUILT_PATTERNS[pattern_name], re.IGNORECASE)


def get_all_pattern_names() -> list[str]:
    """
    Get a list of all available prebuilt pattern names.

    Returns:
        List of pattern names
    """
    return list(PREBUILT_PATTERNS.keys())


# Build category mapping from JSON
PATTERN_CATEGORIES: Final[dict[str, list[str]]] = {}
for pattern_data in _PATTERNS_DATA["patterns"]:
    category = pattern_data["category"]

View on GitHub (pinned to 77b7c6c40c)

Solutions

1. Call get_all_pattern_names() (or read the error message) and use one of the listed names exactly
2. Fix the typo in the rule, e.g. 'creditcard' to 'credit_card'
3. If no prebuilt pattern fits, supply a custom regex in the masking rule instead of a name
4. Verify the patterns.json of your installed LiteLLM version after upgrades and pin the version in deployments

Example fix

# before - unknown name, raises ValueError
rule = {"pattern": "creditcard", "action": "mask"}

# after - exact name from PREBUILT_PATTERNS
from litellm.proxy.guardrails.guardrail_hooks.litellm_content_filter.patterns import get_all_pattern_names
assert "creditcard" in get_all_pattern_names() or True
rule = {"pattern": "credit_card", "action": "mask"}

from litellm.proxy.guardrails.guardrail_hooks.litellm_content_filter.patterns import (  
    get_all_pattern_names,  
    get_compiled_pattern,  
)  
  
valid = set(get_all_pattern_names())  
for rule in masking_rules:  
    if rule.get("prebuilt"):  
        assert rule["pattern"] in valid, f"unknown pattern {rule['pattern']!r}; valid: {sorted(valid)}"  
get_compiled_pattern  # only called after the assertion above

def is_valid_pattern_name(name: object) -> bool:  
    from litellm.proxy.guardrails.guardrail_hooks.litellm_content_filter.patterns import get_all_pattern_names  
    return isinstance(name, str) and name in set(get_all_pattern_names())

When it happens

Trigger: Configuring the content filter guardrail with a pattern name that does not exist in the installed patterns.json: typos ('creditcard' vs 'credit_card'), names from other tools (Presidio detectors), or patterns renamed/removed after a LiteLLM upgrade.

Common situations: Hand-written guardrail YAML referencing pattern names from memory; upgrading LiteLLM where patterns.json changed between versions; configs copied from docs or other repos that drift from the installed version; region-specific names assumed to exist (e.g. 'uk_nino') that were never shipped.

Related errors

• bucket_name must be provided for GCS destination
• MAVVRIK_API_ENDPOINT must be an HTTPS URL
• MAVVRIK_API_ENDPOINT host must be a Mavvrik domain (e.g. htt
• MAVVRIK_API_KEY must be provided for Mavvrik FOCUS destinati
• MAVVRIK_API_ENDPOINT must be provided for Mavvrik FOCUS dest

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/adf8a3e919856265. Report an issue: GitHub.