Zero-Trust Data Sanitization

Developer SDK: Zero-Egress Node.js & Browser PII Sanitization for AI Pipelines

In-process Node.js, WASM & Browser PII sanitization engine. Pre-emptively intercept sensitive customer data and secrets in <0.4ms local RAM before OpenAI, Anthropic, LangChain, or Vector DB ingestion.

100% Local RAM Airplane Mode 0 Bytes Egress
$npm i @privacyscrubber/sdk
Free to start · $299/mo SDK / $2,990/yr Annual
privacyscrubber.com - Developer SDK: Zero-Egress Node.js & Browser PII Sanitization for AI Pipelines
Developer SDK: Zero-Egress Node.js & Browser PII Sanitization for AI Pipelines Interface
Developer SDK & NPM Package

@privacyscrubber/sdk is a PII redaction SDK for Node.js, TypeScript, and the browser. It swaps names, emails, IDs, and API keys for placeholders like [NAME_1] on your own servers in under a millisecond, then puts the real values back into the AI reply. Nothing is sent to PrivacyScrubber or a cloud DLP service, and it ships with 30 industry profiles and drop-in OpenAI and Anthropic wrappers.

Live In-Browser WebAssembly & Node.js Engine

PII Redaction SDK: Live REPL & Latency Benchmark

Execute sub-millisecond in-memory sanitization live in your browser. Verify zero network egress, test bidirectional token detokenization, and inspect the volatile token map in real time across the 5 core scenarios.

0 BYTES EGRESS • AIRPLANE MODE
<0.35ms
Execution Latency
performance.now()
0 Bytes
Network Egress
Local Browser RAM
0 Tokens
Entities Masked
Volatile token dictionary
SOC 2 • PCI DSS • GDPR • HIPAA
Statutory Compliance
Automatic Telemetry
Case Presets:
Active Profile:DevOps & Infrastructure
Raw Payload (Input) 0 chars
100% Volatile RAM Isolationtoken dictionary cleared on tab close
$npm i @privacyscrubber/sdk
Architecture Matrix

The 5 Core Implementation Scenarios: Zero Data Loss Across Modern AI Stacks

From real-time streaming LLM prompts to complex JSON data lakes, tabular CSV files, and multimodal vision payloads.

30 Specialized Profiles
01

Real-Time LLM Prompt & Streaming

OpenAI • Claude • Vercel 3.x

Transparently wrap any standard LLM client in 1 line. Outbound prompts are intercepted and sanitized before cloud egress. The StreamingDetokenizer buffers token boundaries across chunk fragments, restoring original identities in volatile RAM without latency penalties.

// 1-Line Transparent OpenAI / Vercel AI SDK Wrapper
import { wrapOpenAI, wrapAiStream } from'@privacyscrubber/sdk';
const ai = wrapOpenAI(new OpenAI(), { profile: 'legal' });
const res = await ai.chat.completions.create({ model: process.env.OPENAI_MODEL, messages });
Zero context loss • Latency <0.4ms
02

Deep JSON & Vector Lake Ingestion

Pinecone • Qdrant • pgvector

Recursively sanitize arbitrary nested JSON objects, metadata dictionaries, and arrays with sanitizeObject(). Preserves boolean and numeric field types while tokenizing sensitive strings. Eliminates irreversible vector poisoning under GDPR Art. 17 with O(1) constant-time erasure.

// Recursive JSON Sanitization for Vector DB Ingestion
import { sanitizeObject, restoreObject } from'@privacyscrubber/sdk';
const { sanitized, tokenMap } = sanitizeObject(patientRecord, { profile: 'medical' });
await vectorIndex.upsert([{ id: docId, values: embedding, metadata: sanitized }]);
GDPR Art. 17 O(1) Erasure • $0 Re-index
03

Document Pipelines & CSV / Tabular Batch

Pandas • Spark • CSV • S3

Sanitize high-throughput batch files, customer support ticket dumps, and accounting balance sheets without column misalignment or memory spikes. Linear-time ReDoS-hardened state machines process 50MB+ datasets preserving table headers, commas, and formatting boundaries.

// Batch CSV & Tabular Data Stream Sanitization
import { sanitize } from'@privacyscrubber/sdk';
const { scrubbedText, tokenMap } = sanitize(csvContent, { profile: 'finance', preserveStructure: true });
await s3Client.putObject({ Bucket: 'clean-analytics', Body: scrubbedText });
Preserves table structure • PCI DSS v4.0
04

Multimodal Vision & OCR Pre-Prompt Defense

GPT • Claude • OCR

Redact personal identifiers, national IDs (Israel Teudat Zehut, UK NHS, EU IBAN, US SSN), and credentials from OCR text streams before passing image context to vision LLMs. Keeps detected identities out of multimodal training sets and audit caches.

// Multimodal Vision Context Pre-Sanitization
const ocrText = await extractOcrText(scannedDocumentBuffer);
const { scrubbedText, tokenMap } = sanitize(ocrText, { profile: 'general' });
const visionRes = await callVisionModel({ prompt: scrubbedText, image });
Multi-national PII sets • Deterministic entity verification
05

Client-Side Website Form & CRM Interceptor

HTML5 • React • Vue • Zendesk • HubSpot

Intercept user submissions in the browser DOM before network transmission. Sanitizes customer credit cards, emails, phone numbers, and national IDs in local RAM (<0.4ms) with zero cloud egress. Dual-Envelope architecture dispatches clean tokens to public ticketing systems and LLM triage bots while preserving encrypted session maps in your isolated private vault.

// In-Browser DOM Form Interceptor with Dual-Envelope Protection
import { sanitize } from'@privacyscrubber/sdk/browser';
form.addEventListener('submit', async (e) => {
  e.preventDefault();
  const { scrubbedText, tokenMap } = sanitize(textarea.value, { profile: 'support' });
  // 1. Public envelope (safe for Zendesk / Claude triage):
  await fetch('/api/tickets', { method: 'POST', body: JSON.stringify({ message: scrubbedText }) });
  // 2. Private vault envelope (encrypted map for staff resolution):
  await secureVault.store(ticketId, tokenMap);
});
Zero data loss • Defense-in-depth backend fallback • PCI DSS v4.0
Interactive Browser Sandbox (Case 5)Live Website Form & CRM Dual-Envelope Interceptor
Expand Sandbox
Case 05: Client-Side Form & CRM Interceptor

Interactive Website Form Sandbox: Live In-Browser Interception

Test how PrivacyScrubber intercepts customer inquiries, support tickets, and contact form submissions inside the browser DOM before network transmission. Zero PII hits your public CRM or AI models.

DUAL-ENVELOPE VAULT PROTECTION
Sample Customer Support Form DOM Interceptor Active
Client-Side ZTDS Sanitizer
Tokenize in RAM before network POST
<0.3ms In-Memory
POST /api/crm/tickets • Dispatched to Zendesk / Claude Agent 0 BYTES UNMASKED PII
 
100% In-Memory Interception

1. Zero Data Loss (Dual-Envelope)

Public CRM receives tokenized message ([EMAIL_1], [PHONE_1]), while your private vault holds the encrypted tokenMap. Authorized agents click "Reveal" to view real contact info.

Public CRM + VaultZero Context Loss

2. Defense-in-Depth (No Bypass)

Never rely solely on browser JS. The same @privacyscrubber/sdk runs as a 1-line middleware on your Node.js or Next.js backend. If an adversary bypasses the form via cURL, the backend enforces sanitization.

DOM + Node.js APIcURL-Proof Gate

3. Zero Client-Side Secret Leaks

The browser package requires zero private API keys on the frontend. Production quotas and commercial validation are handled at your private backend API gateway ($299/mo Developer SDK license).

Keyless Browser ESMBackend Validated
Enterprise Architecture & Rollout5-Sprint Implementation Roadmap, 4 Strategic Pillars & Data-Flow Lifecycle
Expand Blueprint
Phased Adoption Roadmap

Enterprise Implementation Roadmap: 5 Sprints to Zero-Trust Compliance

Designed for engineering leads, DevSecOps architects, and CISOs deploying generative AI across regulated workflows. A deterministic 5-stage sprint structure that eliminates vendor subprocessor friction, satisfies statutory compliance mandates, and ensures sub-millisecond execution.

Sprint 1 • Day 1 (15 Minutes)Edge & Ingress InterceptionGDPR Art. 4(1) • CCPA • SOC 2 CC6.6

Problem Solved: Prevents accidental leaks of customer names, email addresses, phone numbers, and employee credentials into ChatGPT, Claude, or internal copilot prompts.

Implementation: Drop wrapOpenAI() into backend API gateways or deploy @privacyscrubber/sdk/browser into Next.js App Router client components. Inbound prompts are tokenized before cloud transit; incoming assistant replies are restored deterministically in RAM.

Time-to-Deploy: 15 Mins
Sprint 2 • Week 1Vector Data Lake & RAG DefenseGDPR Art. 17 O(1) Erasure • HIPAA Safe Harbor

Problem Solved: Eliminates irreversible vector poisoning. Embedding raw customer PII bakes personal data permanently into high-dimensional vector graphs, making individual deletion mathematically impossible without expensive full re-indexing.

Implementation: Embed sanitizeObject() into document chunking pipelines before feeding embedding models (Pinecone, Qdrant, Chroma). Store tokenMap in local customer-controlled KMS. Right-to-be-forgotten deletion triggers in O(1) constant time with $0 re-indexing cost.

500k Vector Lake SLA
Sprint 3 • Week 2Enterprise File & Batch IngestionPCI DSS v4.0 Req 3.4 • GLBA • ISO 27001

Problem Solved: Protects tabular financial balance sheets, CSV rosters, and Zendesk/Jira support dumps from exposing credit card PANs, IBANs, and employee salaries to cloud analytics.

Implementation: Run chunked async stream transforms across 30 specialized industry profiles. Structural boundaries (commas, newlines, markdown table cells) remain 100% intact, enabling downstream LLM SQL generators and data analysts to operate without data corruption.

50MB+ Linear Streams
Sprint 4 • Week 3Multimodal Vision & OCR Pre-Flight InterceptionEU AI Act Art. 50 • HIPAA Privacy Rule

Problem Solved: Prevents visual leaks in document image processing. Uploading scanned identity cards, passports, and medical forms to multimodal models (GPT and Claude vision models) bypasses standard text-only DLP filters.

Implementation: Pre-extract text using local offline ZTDS™ OCR WASM; scrub national identifiers (Teudat Zehut, NHS, SIN, TFN) using deterministic structural verification; forward only sanitized bounding box tokens to vision reasoning engines.

Zero Visual PII Leak
Sprint 5 • ContinuousContinuous CI/CD & Production TelemetryNIST SP 800-53 • CMMC 2.0 • FedRAMP High

Problem Solved: Eliminates developer credential drift, uncommitted secrets in Git, and provides verifiable compliance proofs for enterprise security audits without third-party vendor subprocessors.

Implementation: Deploy ps-guard in pre-commit hooks to block raw secrets. Use formatAuditReceipt() to export cryptographic receipts mapping every masked entity to statutory regulations with zero telemetry phone-home calls.

Automated CI/CD Gates
Architectural Mandate

The 4 Strategic Pillars of In-Process Consumer Privacy

Modern SaaS engineering and AI applications bear strict legal and fiduciary responsibility for end-user data under GDPR, CCPA/CPRA, and global privacy statutes. Routing raw customer records through cloud DLP proxies only transfers liability to another third party while introducing latency and fragile network dependencies. The PrivacyScrubber Developer SDK operates as an autonomous consumer data fiduciary directly inside your runtime boundary:

1

Active Consumer Privacy Defense

Systematically enforces privacy rights on behalf of end-users at the application boundary. Sensitive identifiers are protected by default without relying on end-user configuration or manual user opt-ins.

Default Perimeter EnforcementZero User Opt-In
2

Pre-Emptive Interception

Intercepts and quashes customer PII and infrastructure secrets before ingestion into vector databases, logging pipelines, or LLM inference engines—preventing irreversible compliance debt at the perimeter.

Pre-Ingestion SanitizationGDPR Art. 17 Safe
3

Zero Data Loss

Reversible, syntax-preserving deterministic tokens ([NAME_1], [FINANCIAL_1]) preserve 100% LLM reasoning fidelity. Real values are restored exclusively in local RAM with zero loss of downstream analytical utility.

Deterministic Token Mapping100% LLM Fidelity
4

Zero Third Parties

100% in-process execution inside your Node.js or browser memory. Completely eliminates third-party cloud DLP proxies (Skyflow, Lakera, AWS Comprehend, Nightfall), extra vendor DPAs, network hops, and egress fees.

In-Process Node.js / WASM0 Subprocessors

Zero-Trust SDK Architecture: In-Memory Lifecycle

<0.4ms In-Memory Execution

The SDK intercepts raw payloads at your application boundary, tokenizes all PII and infrastructure secrets in volatile RAM, and forwards safe tokens to external LLMs or vector databases. Responses are deterministically restored before reaching the user.

Raw InputPII / Secrets
sanitize()
SDK RAM Engine30 Specialized Profiles
tokenMap
[TOKEN_1]Safe Prompt
OpenAI / LLM
AI ModelTokens Only
restore()
RestoredOriginal Text

Empirical Performance Benchmarks

Node.js 22 LTS • Apple M-Series
0.061ms
Per-Chunk Latency
5,000 char prompt (RAM)
500,000
Vector Benchmark Docs
Sanitized in under 2 mins
$0 vs $14,200
GDPR Erasure Cost
O(1) KMS purge vs re-index

Architecture Comparison: In-Process SDK vs. Cloud DLP Proxies

Why modern engineering organizations choose in-process ZTDS execution over network-bound cloud middleware.

Evaluation FactorLegacy Cloud Proxy DLP (Skyflow, Nightfall) PrivacyScrubber Developer SDK
Execution LocationExternal SaaS cloud proxy (Untrusted network hop)100% In-Process Node.js / Browser RAM (0 egress)
Per-Prompt Latency+250ms to +1,200ms network round-trip<0.4ms linear in-memory execution
GDPR Art. 17 ErasureRequires full vector re-indexing ($14,200+ compute)O(1) constant-time KMS session key purge ($0)
Subprocessor LiabilityAdds vendor to compliance perimeter (6-8 wk review)Zero Subprocessor: No third-party data processor formed
Commercial Pricing$15–$45 / user / mo seat tax + volume egress$299/mo flat for unlimited backend nodes & RAG lakes
ROI & Latency Benchmark Calculator

Calculate Your Monthly Savings & Latency Eliminated

Estimate savings from replacing cloud proxy DLP with local in-memory execution.

Sub-millisecond SLA (<0.4ms)
Select Monthly LLM / RAG Prompts:
User Latency Eliminated
97.2 Hours
350ms saved per outbound request
Cloud Proxy SaaS Cost
$5,000 / mo
$2,500 base + volume egress
Net Annual Savings
$56,412 / yr
94.0% Cost Reduction vs Proxy
PrivacyScrubber: $299/mo flat ($2,990/yr) • Unlimited Nodes • 0 Bytes Egress
1-Line Interceptors • Zero Network Egress

Production Integration Snippets: OpenAI, Anthropic, LangChain & LlamaIndex

Drop-in client wrappers and middleware interceptors. Sub-millisecond in-memory execution (<0.4ms) with zero third-party subprocessors.

Flat $299/mo • Unlimited Nodes
openai-interceptor.ts1-Line Transparent Drop-In
// 1. Install zero-dependency SDK: npm i @privacyscrubber/sdk openai
import OpenAI from 'openai';
import { wrapOpenAI } from '@privacyscrubber/sdk';

// 2. Wrap client: In-process consumer data fiduciary (<0.4ms RAM execution)
// Flat $299/mo SDK license: Unlimited production microservices & 0 bytes cloud egress
const openai = wrapOpenAI(new OpenAI({ apiKey: process.env.OPENAI_API_KEY }), {
  profile: 'legal',          // 30 industry profiles (dev, medical, finance, etc.)
  detectSecrets: true,       // Intercepts AWS keys, Bearer JWTs, and database URIs
  licenseKey: process.env.PRIVACYSCRUBBER_KEY // Air-gapped Ed25519 (0 phone-home pings)
});

// 3. Standard OpenAI syntax: prompts are intercepted & tokenized in local memory before egress.
// Responses are automatically detokenized in RAM before returning to your caller.
const stream = await openai.chat.completions.create({
  model: process.env.OPENAI_MODEL,
  messages: [{
    role: 'user',
    content: 'Review contract for Johnathan Vance (SSN: 219-44-8891, email: jvance@acme.corp).'
  }],
  stream: true // StreamingDetokenizer buffers token boundaries across chunk fragments
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content || '');
}
0 Code Changes NeededStream Buffer Detokenizer IncludedHIPAA • GDPR • SOC 2
Unlimited Internal Nodes • $299/mo Flat
Deploy across Fastify, Express, Rust, Python, and Node.js microservices with zero external dependencies.
Procurement & Compliance Clearance48-Hour InfoSec Sign-Off, Dual-Key Architecture & A4 Conformance Certificate
Expand Procurement Pack
Enterprise AI Sales Velocity & Procurement Clearance

From 6-Month InfoSec Review to 48-Hour Sign-Off

Enterprise buyers refuse to send confidential records and PII to frontier AI clouds. Combining in-process @privacyscrubber/sdk tokenization with public ZTDS.ai accreditation turns compliance from a sales blocker into an unfair conversion advantage.

ZTDS RFC v1.0 CERTIFIED
01
AI SaaS & Integrators

Unlock Enterprise Deals

Shorten 3–6 month vendor risk reviews (SIG Lite, CAIQ, SOC 2 questionnaires). Provide the official A4 Conformance Certificate and verified registry listing proving zero customer PII leaves your infrastructure.

48-Hour Procurement Clearance
02
CISOs & Legal DPOs

Statutory Risk Safe Harbor

Technical & organizational measures (TOMs) under GDPR Art. 32 and Israel PPL Amendment 13 Sec. 17. RAM-only sanitization physically prevents direct PII egress and eliminates officer personal administrative liability.

Personal DPO & CISO Shield
03
Legal, Tax & Advisory

Zero-Risk AI Adoption

Query ChatGPT, Claude, and Gemini with confidential client records, contracts, and financial workpapers. Work safely in 100% Airplane Mode with 1-click deterministic reverse token re-hydration in local RAM.

1-Click Reverse De-Tokenization

The Dual-Key System: Technical Runtime vs. Public Trust

Architecture Invariant

Subscribers receive two complementary cryptographic artifacts: a private in-process runtime key that unlocks unlimited throughput on your machines, and a public accreditation profile on ZTDS.ai that verifies compliance to prospective enterprise buyers.

1. Technical License Keyprocess.env.PS_LICENSE_KEY

In-Process Runtime Engine Key

  • Format:PS-SDK-[HEX]-[CHECKSUM] (Ed25519 token)
  • Execution: 100% offline in RAM. Zero phone-home calls or external pings
  • Capability: Unlocks unlimited characters and all 30 industry profiles
Private to your internal backend nodes • <0.4ms latency
2. Public Accreditationztds.ai/verify/?target=...

ZTDS Verified Registry Listing & A4 Cert

  • Format: Public registry URL + printable A4 PDF Certificate of Conformance
  • Verification: WebCrypto Ed25519 cryptographic signature verified by client CISOs
  • Function: Ingested into Vanta, Drata, and enterprise security questionnaires
Public proof of zero network egress • Collapses procurement
Official Audit Artifact

Official A4 Certificate of Conformance

Every production SDK deployment produces legally authoritative documentation. Export vector-crisp, printable A4 PDF certificates signed under the Zero-Trust Data Sanitization standard (ZTDS RFC v1.0) to satisfy enterprise customer security reviews, DPIAs, and statutory audits.

RFC v1.0 Valid
ZTDS CONFORMANCE
Standard Seal
Egress Guarantee:0 BYTES (RAM-ONLY)
Legal Standard:GDPR Art. 32 / TOMs
Audit Receipt:SHA-256 Verified
Attestation:Ed25519 Signed
Printable A4 PDF Verify →
0-Subprocessor Procurement Justification

Ready for Immediate CISO & DPO Approval

Because the SDK executes 100% inside your local application runtime and transmits 0 bytes to PrivacyScrubber or third parties, PrivacyScrubber never receives your data, which typically keeps it outside GDPR Article 28 processor and HIPAA BAA scope (your counsel makes the final call). Review our pre-filled 40-Question Vendor Security Assessment (SIG Lite / CAIQ) and statutory Zero-Processor Safe Harbor Technical Pack. Your security team can verify zero network sockets using Wireshark or DevTools in under 60 seconds.

40-Point SIG Lite Matrix
Technical & IT Security Summary
Client-Side Only

"Air-gapped Developer SDK for Node.js and WASM. Intercept PII in local RAM before LLM ingestion. Helps meet HIPAA Safe Harbor and SOC 2 controls."

Install via npm: Add @privacyscrubber/sdk to your project using npm, yarn, or pnpm. Operates with zero runtime dependencies across Node.js 18+ and modern browser ESM contexts.
Wrap LLM or Sanitize Direct: Use wrapOpenAI(client) for 1-line transparent prompt sanitization, or call sanitize(text, { profile: 'dev' }) to strip PII and infrastructure secrets before vector DB ingestion.
Deterministic Restore & Audit Receipts: Re-hydrate tokens in AI responses with restore(response, tokenMap) and output signed Markdown/JSON CISO audit receipts mapped to GDPR, HIPAA, and SOC 2.

Embeddable Node/WASM Guard for Developer SDK

Deploy reversible @privacyscrubber/sdk masking inside Node.js, TypeScript, and RAG chunkers at zero egress cost.

ZTDS™ RFC v1.0 Verified Capability·ZTDS-CERT-2026-PRIVACYSCRUBBER-WEB-D02DC6
Verify Cryptographic Seal

Frequently Asked Questions

Technical Details & Privacy Architecture

Can I embed the SDK directly into my website's HTML contact or support form?
Yes. Importing @privacyscrubber/sdk/browser intercepts HTML form submissions inside local browser RAM in under 0.4ms before network requests leave the device. Credit cards, emails, phones, and IDs tokenize automatically so clean payloads reach Zendesk or HubSpot while encrypted maps seal separately in your private vault.
What environments and runtimes does @privacyscrubber/sdk support?
PrivacyScrubber SDK is completely isomorphic with zero runtime dependencies, running natively across Node.js 18+, Bun, Deno, Next.js Server Actions, Express, Fastify, AWS Lambda, Cloudflare Workers, Electron, and Tauri. Browser builds integrate directly with React, Vue, Svelte, and Web Workers in local memory.
How does the 1-line wrapOpenAI helper work?
The wrapOpenAI wrapper intercepts any standard OpenAI client instance in-process before prompts traverse the network. It scans and replaces sensitive PII and infrastructure secrets with reversible surrogate tokens in local RAM, sends tokenized text to OpenAI, and transparently restores original values in returned completions.
How does the SDK protect Vector Databases (Pinecone, Chroma, Qdrant, Milvus, pgvector)?
Pre-embedding sanitization strips raw PII in memory in under 0.065ms before chunks reach Pinecone, Qdrant, Chroma, Milvus, or pgvector. Storing surrogate tokens instead of cleartext identities eliminates irreversible vector poisoning under GDPR Article 17, enabling constant-time O(1) deletion via key revocation without $14,200 index rebuilds.
How does the SDK handle multi-turn AI agents and state consistency?
Autonomous frameworks such as LangChain, AutoGen, CrewAI, and LlamaIndex instantiate PrivacyScrubberEngine to keep stateful token dictionaries inside local RAM across multi-turn conversations. Identical identifiers resolve deterministically as consistent tokens from Turn 1 through Turn 10 while supporting markFalsePositive and resetSession lifecycle methods.
Does the SDK detect infrastructure secrets and DevOps tokens?
Yes. Enabling detectSecrets scans source code, configuration dumps, and telemetry for AWS access keys, Bearer JWTs, GitHub tokens, Stripe live secret keys, OpenAI credentials, and database URIs with passwords. All sensitive secrets become surrogate tokens inside local RAM before any LLM call occurs.
What compliance frameworks and audit telemetry are generated?
Every sanitize call calculates deterministic risk scores and tamper-evident SHA-256 Markdown audit receipts inside local memory with zero external API calls. Entities map automatically across GDPR Article 4(1), HIPAA Safe Harbor section 164.514, SOC 2 Type II, PCI DSS v4.0, and NIST SP 800-53 controls.
Does the SDK require an outbound internet connection to validate licenses?
No. PrivacyScrubber SDK operates 100% air-gapped with zero phone-home pings, zero user telemetry, and zero external license validation endpoints. Commercial PRO, TEAMS, SDK, OEM, and Enterprise keys verify mathematically inside process memory using offline Ed25519 cryptographic checksums and asymmetric RS256 expiration checks.
How does the SDK compare to MS Presidio and Google Cloud DLP?
Google Cloud DLP and AWS Comprehend transmit raw cleartext across public networks, adding 150-400ms latency and third-party subprocessor DPAs, while MS Presidio requires heavy 500MB Python containers. PrivacyScrubber SDK runs as a zero-dependency 180KB JavaScript module executing locally under one millisecond with zero egress.
Is the SDK fully typed with TypeScript?
Yes. Every release ships strict bundled index.d.ts declarations requiring zero external @types packages. Methods including sanitize, restore, wrapOpenAI, createLangChainTransform, and PrivacyScrubberEngine expose complete IntelliSense types for configuration flags, all 30 specialized vertical profiles, and compliance telemetry objects inside VS Code and Cursor.
How does pricing and quota management work for the Developer SDK, and is it free to try?
Yes. Local evaluation is permanently free with a built-in session quota of 15,000 General characters and 5,000 specialized characters across 30 vertical profiles, requiring zero accounts or credit cards. Production nodes run on a flat $299/month or $2,990/year SDK license; 7-day corporate TEAMS pilots expire fail-closed.
What is the difference between the technical SDK license key and the public ZTDS Certification listing?
The offline Ed25519 SDK license key (PS-SDK-...) configures local process execution to unlock unlimited production characters across all 30 profiles with zero network pings. Conversely, the public ZTDS Certification listing on ztds.ai provides independent conformance proof that compresses enterprise vendor security reviews into rapid approval.
Pre-Commit & CI/CD Zero-Trust Shield

Git Pre-Commit Leak Shield & 7-Day Evaluation Trial

Block leaked cloud credentials, API keys, database connection strings, and consumer PII before git commits leave developer workstations. Plus instant offline cryptographic evaluation trials.

<5ms HOOK LATENCY • 0 EGRESS
Git Pre-Commit Hook (1-Click Setup) Binary: ps-git-guard

Install the zero-dependency hook into .git/hooks/pre-commit. Scans staged diffs in <5ms and aborts commits if unmasked secrets are detected:

# 1. Install hook in active repo:
npx ps-git-guard install
# 2. Or scan staged changes anytime:
npx ps-git-guard
Interception Alert: Blocked commit: detected 2 sensitive token(s) (AWS_KEY, OPENAI_KEY) in staged git diff.
Uninstall anytime: npx ps-git-guard uninstall Zero Config
Instant 7-Day PoC / Staging Evaluation Offline Time-Bomb

Evaluating in enterprise staging, ETL test environments, or PoC benchmarks? Generate a standalone cryptographic evaluation trial key directly from your terminal:

# Activate 7-day evaluation key in terminal:
npx @privacyscrubber/sdk trial
100% AIR-GAPPED
No email or credit card needed
ALL 30 PROFILES
Unlimited chars for 7 days
Zero-Latency In-Memory Execution

Deploy Zero-Trust PII Sanitization in Under 30 Seconds

Free with built-in quota for local dev & prototyping. Zero credit card. Zero account sign-up. Commercial license covers unlimited internal production nodes and RAG lakes.

$npm i @privacyscrubber/sdk
Free Forever with Quota 14-Day Money-Back Guarantee Air-Gapped Ed25519 VerifiedCompare Pricing Plans →