Bring Ai Guardrails
to Vercel AI SDK

Analyze and redact content in LLM inputs/outputs

Scan text for PII, malicious content, and prompt injections

Guard LLM chat completions with integrated logging and tracing

Create a single secure audit log entry

Create multiple secure audit log entries

Search the audit log using natural language queries

Paginate through audit search results

Finalize the flow and receive session tokens

Start a sign-up or sign-in flow

Update flow state (e.g., submit password, OTP)

List active user sessions

Invalidate sessions

Programmatically create a user

Check if a subject has permission for an action on a resource

List all resources a subject can access

Define relationships for AuthZ

Get WHOIS details for a domain

Check if an IP originates from an embargoed country

Check a 2-character ISO country code against embargo lists

Scan a file for malware

Retrieve reputation scores for domains, URLs, or file hashes

Retrieve location data for an IP

Detect if an IP is a proxy

Get reputation score and verdict for an IP

Detect if an IP is a VPN

Check if a password hash prefix appears in breach data

Redact specific fields in a JSON object using JSONPath

Redact sensitive data from plain text

Apply sanitization rules to a file and receive a cleaned version

Delete files or folders

Create folders in Secure Share

Download a file or retrieve metadata

List objects in a bucket or folder with filtering

Decrypt data previously redacted using FPE

Check if an email, phone, or username appears in known data breaches

Perform cryptographic decryption

Perform cryptographic encryption

Retrieve item details or secret values from Vault

Generate symmetric or asymmetric keys in Vault

Store a secret or token in Vault