@privacyscrubber/sdk 2.4.0 → 2.4.1

package/benchmark.js

@@ -114,17 +114,8 @@ function benchmarkWorkload(label, targetBytes, warmupCount, measurementCount, pr
     const rawPayload = generateRealisticWorkload(targetBytes);
     const actualBytes = Buffer.byteLength(rawPayload, 'utf8');
 
-    // Generate valid test license key to avoid community tier truncation
-    const salt = "ZTDS_SALT_2026_!@#";
-    const core = "PROD9876KEY5";
-    const tier = "SDK";
-    const inputStr = core + salt + tier;
-    let sum = 0;
-    for (let j = 0; j < inputStr.length; j++) {
-        sum += inputStr.charCodeAt(j) * (j + 1);
-    }
-    const checksum = (sum % 9999).toString().padStart(4, '0');
-    const licenseKey = `PS-${tier}-${core}-${checksum}`;
+    // Optional commercial evaluation license key from environment (defaults to Community tier)
+    const licenseKey = process.env.PRIVACYSCRUBBER_KEY || undefined;
 
     // Warm-up phase (V8 TurboFan JIT optimization)
     let warmupResult;
@@ -198,7 +189,15 @@ export function runBenchmarks(options = {}) {
     const profile = options.profile || REQUESTED_PROFILE;
     const itersOverride = options.iterations || ITERATIONS_OVERRIDE;
 
+    const hasCommercialKey = Boolean(process.env.PRIVACYSCRUBBER_KEY);
     const suites = [
+        {
+            label: '1 KB (Short Agentic Tool Call / Chat Turn)',
+            bytes: 1 * 1024,
+            warmup: 50,
+            iterations: itersOverride > 0 ? itersOverride : 500,
+            targetSlaMs: 0.8
+        },
         {
             label: '2 KB (Interactive LLM Chat Prompt)',
             bytes: 2 * 1024,
@@ -206,6 +205,13 @@ export function runBenchmarks(options = {}) {
             iterations: itersOverride > 0 ? itersOverride : 500,
             targetSlaMs: 1.0 // Sub-millisecond ZTDS SLA invariant
         },
+        {
+            label: '4 KB (Multi-Entity Solid Pod / Clinical Note)',
+            bytes: 4 * 1024,
+            warmup: 30,
+            iterations: itersOverride > 0 ? itersOverride : 300,
+            targetSlaMs: 2.5
+        },
         {
             label: '10 KB (Enterprise Support Ticket / Incident)',
             bytes: 10 * 1024,
@@ -213,20 +219,22 @@ export function runBenchmarks(options = {}) {
             iterations: itersOverride > 0 ? itersOverride : 250,
             targetSlaMs: 5.0
         },
-        {
-            label: '50 KB (RAG Context Chunk / Knowledge Retrieval)',
-            bytes: 50 * 1024,
-            warmup: 15,
-            iterations: itersOverride > 0 ? itersOverride : 100,
-            targetSlaMs: 25.0
-        },
-        {
-            label: '250 KB (Batch Processing / Multi-Page Document)',
-            bytes: 250 * 1024,
-            warmup: 5,
-            iterations: itersOverride > 0 ? itersOverride : 30,
-            targetSlaMs: 150.0
-        }
+        ...(hasCommercialKey ? [
+            {
+                label: '50 KB (RAG Context Chunk / Knowledge Retrieval)',
+                bytes: 50 * 1024,
+                warmup: 15,
+                iterations: itersOverride > 0 ? itersOverride : 100,
+                targetSlaMs: 25.0
+            },
+            {
+                label: '250 KB (Batch Processing / Multi-Page Document)',
+                bytes: 250 * 1024,
+                warmup: 5,
+                iterations: itersOverride > 0 ? itersOverride : 30,
+                targetSlaMs: 150.0
+            }
+        ] : [])
     ];
 
     const results = [];

package/browser.js

@@ -28,16 +28,33 @@ if (scrubberPkg && typeof scrubberPkg.init === 'function') {
  * Built-in patterns for DevOps & Cloud Secrets detection.
  */
 export const DEVOPS_SECRETS_DETECTOR = [
-    { name: 'AWS Credentials', regex: /\b(?:AKIA|ASIA|AGPA|AIDA|AROA|AIPA)[A-Z0-9]{16}\b/g, label: 'AWS_KEY' },
+    { name: 'AWS Credentials', regex: /\b(?:AKIA|ASIA|AGPA|AIDA|AROA|AIPA)[A-Z0-9]{12,28}\b/g, label: 'AWS_KEY' },
+    { name: 'Google Cloud / Gemini API Key', regex: /\bAIza[0-9A-Za-z_-]{35}\b/g, label: 'API_TOKEN' },
+    { name: 'Anthropic API Key', regex: /\bsk-ant-(?:api\d{2}|admin\d{2}|sid\d{2})-[a-zA-Z0-9_-]{16,128}\b/g, label: 'API_TOKEN' },
+    { name: 'HuggingFace / Perplexity API Key', regex: /\b(?:hf_[a-zA-Z0-9]{20,64}|pplx-[a-zA-Z0-9]{24,64})\b/g, label: 'API_TOKEN' },
+    { name: 'SendGrid API Key', regex: /\bSG\.[a-zA-Z0-9_-]{16,32}\.[a-zA-Z0-9_-]{32,64}\b/g, label: 'API_TOKEN' },
+    { name: 'Postman API Key', regex: /\bPMAK-[a-fA-F0-9]{24}-[a-fA-F0-9]{34}\b/g, label: 'API_TOKEN' },
+    { name: 'GitHub Fine-Grained PAT', regex: /\bgithub_pat_[a-zA-Z0-9]{20,30}_[a-zA-Z0-9]{40,80}\b/g, label: 'API_TOKEN' },
+    { name: 'Slack Enterprise / Refresh Token', regex: /\bxoxe\.xox[baprs]-\d-[a-zA-Z0-9-]{10,}\b|\bxox[eapbrs]-[a-zA-Z0-9-]{10,}\b/g, label: 'API_TOKEN' },
+    { name: 'Telegram Bot API Token', regex: /\b\d{8,10}:AA[A-Za-z0-9_-]{33}\b/g, label: 'API_TOKEN' },
+    { name: 'Age Encryption Secret Key', regex: /\bAGE-SECRET-KEY-1[A-Z0-9]{58}\b/g, label: 'SECRET' },
+    { name: 'Modern Cloud & AI Platform Key', regex: /\b(?:sbp_[a-fA-F0-9]{40}|dop_v1_[a-fA-F0-9]{64}|doo_v1_[a-fA-F0-9]{64}|dp\.(?:st|pt|ct|sa)\.[a-zA-Z0-9_-]{20,}|lin_api_[a-zA-Z0-9]{32,64}|ntn_[a-zA-Z0-9]{32,64}|secret_[a-zA-Z0-9]{32,64}|gsk_[a-zA-Z0-9]{32,64}|r8_[a-zA-Z0-9]{24,64}|tvly-[a-zA-Z0-9_-]{20,64}|fc-[a-fA-F0-9]{32}|dapi[a-fA-F0-9]{32}(?:-\d)?|re_[a-zA-Z0-9_]{20,64}|napi_[a-zA-Z0-9]{32,64}|[a-zA-Z0-9]{14}\.atlasv1\.[a-zA-Z0-9]{50,80}|pypi-[A-Za-z0-9_-]{16,128}|whsec_[A-Za-z0-9+/=_-]{24,128}|shpat_[a-fA-F0-9]{32}|shpss_[a-fA-F0-9]{32}|shpca_[a-fA-F0-9]{32}|shppa_[a-fA-F0-9]{32}|hvs\.[A-Za-z0-9_-]{24,128}|hvb\.[A-Za-z0-9_-]{24,128}|glsa_[A-Za-z0-9]{32}_[A-Fa-f0-9]{8}|glc_[A-Za-z0-9+/=_-]{20,128}|pscale_tkn_[A-Za-z0-9_.-]{20,64}|pscale_pw_[A-Za-z0-9_.-]{20,64}|pat[A-Za-z0-9]{14}\.[a-fA-F0-9]{64}|pat-(?:na1|eu1)-[a-fA-F0-9]{8}-[a-fA-F0-9]{4}-[a-fA-F0-9]{4}-[a-fA-F0-9]{4}-[a-fA-F0-9]{12}|sq0atp-[A-Za-z0-9_-]{22,64}|sq0csp-[A-Za-z0-9_-]{32,64}|NRAK-[A-Z0-9]{27}|v1\.0-[a-fA-F0-9]{24}-[a-fA-F0-9]{16,160}|ATATT3xFfGF0[A-Za-z0-9+/=_-]{24,128}|tskey-(?:auth|api|client)-[A-Za-z0-9_-]{20,96}|sntry[su]_[A-Za-z0-9+/=_-]{32,160}|(?:pk|sk)\.eyJ1[A-Za-z0-9_-]{20,}\.[A-Za-z0-9_-]{16,}|key-[a-f0-9]{32}|[a-f0-9]{32}-us\d{1,2})\b/g, label: 'API_TOKEN' },
+    { name: 'Twilio Account SID / API Key', regex: /\b(?:AC|SK)[a-fA-F0-9]{32}\b/g, label: 'API_TOKEN' },
+    { name: 'Bcrypt / Argon2 Password Hash', regex: /(?<![A-Za-z0-9])(?:\$2[aby]\$\d{2}\$[./A-Za-z0-9=]{20,60}|\$argon2(?:id|i|d)\$[^\s"']{15,120})/g, label: 'SECRET' },
     { name: 'JSON Web Token (JWT)', regex: /\beyJ[a-zA-Z0-9_-]+\.[a-zA-Z0-9_-]+\.[a-zA-Z0-9_-]+\b/g, label: 'JWT_TOKEN' },
-    { name: 'API Token/Key (GitHub/Slack/NPM)', regex: /\b(?:ghp|gho|ghu|ghs|ghr|glpat|npm|xox[baprs])[-_][A-Za-z0-9_]{10,}\b/g, label: 'API_TOKEN' },
-    { name: 'Stripe API Key', regex: /\b(?:[rs]k)_(?:test|live)_[a-zA-Z0-9]{24,}\b/g, label: 'STRIPE_KEY' },
+    { name: 'API Token/Key (GitHub/Slack/NPM)', regex: /\b(?:ghp|gho|ghu|ghs|ghr|glpat|npm|xox[baprs])[-_][A-Za-z0-9_-]{10,}\b/g, label: 'API_TOKEN' },
+    { name: 'Stripe API Key', regex: /\b(?:[rs]k)_(?:test|live)_[a-zA-Z0-9]{10,}\b/g, label: 'STRIPE_KEY' },
     { name: 'OpenAI Project API Key', regex: /\b(?:sk|pk)-(?:proj-)?[a-zA-Z0-9_-]{16,}\b/gi, label: 'OPENAI_KEY' },
+    { name: 'Slack / Discord Incoming Webhook URL', regex: /\bhttps:\/\/(?:hooks\.slack\.com\/services\/T[A-Z0-9_]+\/B[A-Z0-9_]+\/[A-Za-z0-9_-]+|(?:ptb\.|canary\.)?discord(?:app)?\.com\/api\/webhooks\/\d{17,20}\/[A-Za-z0-9_-]{60,70})\b/gi, label: 'SECRET' },
+    { name: 'Azure Connection String / SAS Token', regex: /(?:DefaultEndpointsProtocol=https?;AccountName=[^;\s"']+;AccountKey=[A-Za-z0-9+/=]{20,}(?:;EndpointSuffix=[^;\s"']+)?|Endpoint=sb:\/\/[^;\s"']+;SharedAccessKeyName=[^;\s"']+;SharedAccessKey=[A-Za-z0-9+/=]{16,}|\bSharedAccessSignature=sr=[^\s"']+&sig=[A-Za-z0-9%/+=]{16,}|\bAccountKey=[A-Za-z0-9+/=]{40,88}==)/gi, label: 'SECRET' },
     { name: 'Database Connection URI', regex: /\b(?:postgres(?:ql)?|mysql|mongodb(?:\+srv)?|redis|amqp|mssql):\/\/[^\s"']+/gi, label: 'DB_URI' },
-    { name: 'Embedded Credentials URI', regex: /\b(?:https?|s?ftp):\/\/[a-zA-Z0-9_.%+-]+:[^\s/@:"']+@[a-zA-Z0-9_.-]+(?::\d+)?(?:\/[^\s"']*)?/gi, label: 'SECRET' },
+    { name: 'Embedded Credentials URI', regex: /\b(?:https?|hxxps?|s?ftp|fxp|wss?):\/\/[a-zA-Z0-9_.%+-]+:[^\s/@:"']+@[a-zA-Z0-9_.-]+(?:\[\.\][a-zA-Z0-9_.-]+|\(\.\)[a-zA-Z0-9_.-]+)*(?::\d+)?(?:\/[^\s"']*)?/gi, label: 'SECRET' },
     { name: 'URL Query Parameter Secret', regex: /(?<=[?&](?:api_key|apikey|access_token|client_secret|auth_token|token|secret|password|private_key)=)[^\s"'\x26#)>]{8,}/gi, label: 'SECRET' },
+    { name: 'CLI Credential Flag / Curl User Auth', regex: /(?:\bcurl\b[^\r\n]*?\s(?:-u|--user)\s+["']?[A-Za-z0-9_.+-]+:[^\s"']+|\b(?:--password|--api-key|--apikey|--access-token|--auth-token|--client-secret|--secret-key|--private-key)(?:=|\s+)["']?[^\s"']{6,}["']?)/gi, label: 'SECRET' },
+    { name: 'HTTP API Key Header', regex: /\b(?:X-API-Key|X-Amz-Security-Token|X-Functions-Key|X-Shopify-Access-Token|X-Auth-Token|Api-Key)\s*:\s*["']?[A-Za-z0-9_.\-~+/=]{12,}["']?/gi, label: 'SECRET' },
+    { name: 'Docker Registry Auth / K8s Secret Base64', regex: /(?<=["']?(?:auth|identitytoken|registrytoken|\.dockerconfigjson)["']?\s*:\s*["'])[A-Za-z0-9+/=]{16,}(?=["'])/gi, label: 'SECRET' },
     { name: 'Bearer / Auth Header Token', regex: /(?<=\b(?:Bearer|Token)\s+)[a-zA-Z0-9_\-\.~+/=]{16,}\b/gi, label: 'SECRET' },
-    { name: 'Database/API Secret', regex: /\b(DB|POSTGRES|REDIS|MYSQL|AWS|SECRET|PASSWORD|TOKEN|API|KEY)[A-Z0-9_]*\s*[:=]\s*[^ \t\r\n"']{8,}\b/gi, label: 'SECRET' },
+    { name: 'Database/API Secret', regex: /\b(DB|POSTGRES|REDIS|MYSQL|AWS|SECRET|PASSWORD|TOKEN|API|KEY)[A-Z0-9_]*\s*[:=]\s*[^ \t\r\n"']{8,}/gi, label: 'SECRET' },
     { name: 'Private Cryptographic Key (PEM Block)', regex: /-----BEGIN (?:RSA |EC |PGP |DSA |OPENSSH |ENCRYPTED )?PRIVATE KEY(?: BLOCK)?-----[\s\S]{10,8192}?-----END (?:RSA |EC |PGP |DSA |OPENSSH |ENCRYPTED )?PRIVATE KEY(?: BLOCK)?-----/g, label: 'SECRET' },
     { name: 'Private Cryptographic Key (Header)', regex: /-----BEGIN (?:RSA |EC |PGP |DSA |OPENSSH |ENCRYPTED )?PRIVATE KEY(?: BLOCK)?-----/g, label: 'SECRET' }
 ];
@@ -147,7 +164,7 @@ export function buildAuditTelemetry(currentTokenMap = {}) {
     }
 
     const ipHoldings = {
-        standard: 'IETF draft-sibiryakov-ztds-protocol-00',
+        standard: 'IETF draft-sibiryakov-ztds-protocol-01',
         standardUrl: 'https://datatracker.ietf.org/doc/draft-sibiryakov-ztds-protocol/',
         patent: "U.S. & Int'l Patents Pending (Paris Convention Priority)",
         trademark: 'ZTDS™ Reg. #182655957 (ILPO Cl 9 & 42)'
@@ -157,27 +174,27 @@ export function buildAuditTelemetry(currentTokenMap = {}) {
         {
             statute: 'EU AI Act (Art. 50)',
             scope: 'AI Transparency & Data Minimization',
-            rule: '0-byte external network transmission; prevents unauthorized ingestion into frontier models.'
+            rule: 'Sanitization runs locally by design; reduces the content sent to external AI models. Informational mapping, not a compliance determination.'
         },
         {
             statute: 'GDPR (Art. 4, 25, 28, 32)',
             scope: 'Pseudonymization & Processor Liability',
-            rule: 'Volatile client-side tokenization eliminates GDPR Article 28 data processor liability.'
+            rule: 'Client-side pseudonymization is a technical measure supporting Art. 25 and 32. It does not by itself remove controller or processor obligations.'
         },
         {
             statute: 'HIPAA 45 CFR § 164.514(b)',
             scope: 'Safe Harbor De-Identification',
-            rule: 'Deterministic removal of all 18 PHI identifiers prior to AI analysis.'
+            rule: 'Pattern-based masking of direct identifiers supports Safe Harbor de-identification workflows. Review output for residual identifiers before sharing.'
         },
         {
             statute: 'IRC § 7216 & IRS Pub. 1075',
             scope: 'Tax Return Information Confidentiality',
-            rule: 'Strict criminal penalty safeguard against unauthorized disclosure of tax records to cloud LLMs.'
+            rule: 'Masking identifiers before AI use can reduce exposure of tax return information. Informational mapping, not legal advice.'
         },
         {
             statute: 'FRE 502',
             scope: 'Attorney-Client Privilege Preservation',
-            rule: 'Automated in-memory redaction prevents inadvertent waiver of legal privilege in AI workflows.'
+            rule: 'In-memory redaction can reduce disclosure of privileged details in AI workflows. Privilege determinations remain a legal judgment.'
         }
     ];
 
@@ -545,6 +562,10 @@ export class PrivacyScrubberEngine {
             delete this.sessionMap[key];
         }
     }
+
+    createStreamRestorer(options = {}) {
+        return createStreamTokenRestorer(this.sessionMap, options);
+    }
 }
 
 /**
@@ -576,13 +597,13 @@ export function buildStreamReplacer(tokenMap = {}, options = {}) {
 
     for (const k of sortedKeys) {
         const val = tokenMap[k];
-        const inner = k.replace(/^\[|<|\{\{|__|\]|>|\}\}|__/g, '');
+        const inner = k.replace(/^\[|<|\{\{?|__|\]|>|\}\}?|__/g, '');
         const match = inner.match(/^([A-Za-z_0-9]+?)[-_ ]?(\d+)$/);
 
         if (match) {
-            const label = escapeRegex(match[1]);
+            const label = escapeRegex(match[1]).replace(/[_-]/g, '\\\\?[_-]');
             const idx = match[2];
-            const patternStr = '(?:\\[\\s*' + label + '[-_\\s]*0*' + idx + '\\s*\\]|<\\s*' + label + '[-_\\s]*0*' + idx + '\\s*>|\\{\\{\\s*' + label + '[-_\\s]*0*' + idx + '\\s*\\}\\}|__\\s*' + label + '[-_\\s]*0*' + idx + '\\s*__|\\b' + label + '[-_\\s]+0*' + idx + '\\b)';
+            const patternStr = '(?:(?:\\[|\\\\u005[Bb]|%5[Bb]|&(?:amp;)?#0*91;|&(?:amp;)?#[xX]0*5[Bb];|&(?:amp;)?lbrack;|【|［|〔|「|『)\\s*[*`~]*' + label + '[-_\\\\\\s]*0*' + idx + '[*`~]*[.,;:!?]?\\s*(?:\\]|\\\\u005[Dd]|%5[Dd]|&(?:amp;)?#0*93;|&(?:amp;)?#[xX]0*5[Dd];|&(?:amp;)?rbrack;|】|］|〕|」|』)|<\\s*' + label + '[-_\\\\\\s]*0*' + idx + '[.,;:!?]?\\s*>|\\{\\{?\\s*' + label + '[-_\\\\\\s]*0*' + idx + '[.,;:!?]?\\s*\\}\\}?|__\\s*' + label + '[-_\\\\\\s]*0*' + idx + '\\s*__|\\b' + label + '[-_\\s]+0*' + idx + '\\b)';
             patterns.push(patternStr);
             const normKey = (match[1] + '_' + parseInt(idx, 10)).toUpperCase();
             lookup.set(normKey, val);
@@ -611,10 +632,14 @@ export function buildStreamReplacer(tokenMap = {}, options = {}) {
             } else if (lookup.has(coreMatch.toUpperCase())) {
                 restored = lookup.get(coreMatch.toUpperCase());
             } else {
-                const inner = coreMatch.replace(/^\[|<|\{\{|__|\]|>|\}\}|__/g, '').trim();
-                const m = inner.match(/^([A-Za-z_0-9]+?)[-_ ]?(\d+)$/);
+                const inner = coreMatch
+                    .replace(/^(?:\[|\\u005[Bb]|%5[Bb]|&(?:amp;)?#0*91;|&(?:amp;)?#[xX]0*5[Bb];|&(?:amp;)?lbrack;|【|［|〔|「|『|<|\{\{?|__)|(?:\]|\\u005[Dd]|%5[Dd]|&(?:amp;)?#0*93;|&(?:amp;)?#[xX]0*5[Dd];|&(?:amp;)?rbrack;|】|］|〕|」|』|>|\}\}?|__)$/gi, '')
+                    .replace(/^[*`~\s]+|[.,;:!?*`~\s]+$/g, '')
+                    .trim();
+                const m = inner.match(/^([A-Za-z_0-9]+?)[-_\\\s]*(\d+)$/);
                 if (m) {
-                    const normKey = (m[1] + '_' + parseInt(m[2], 10)).toUpperCase();
+                    const cleanLabel = m[1].replace(/\\([_-])/g, '$1');
+                    const normKey = (cleanLabel + '_' + parseInt(m[2], 10)).toUpperCase();
                     if (lookup.has(normKey)) {
                         restored = lookup.get(normKey);
                     }
@@ -720,18 +745,17 @@ export class StreamingDetokenizer {
         this.pendingBuffer += text;
 
         // Candidate unclosed token at tail of buffer (inspect tail window up to 160 chars)
-        // Matches incomplete tokens starting with [, <, {{, or __, plus single delimiter prefixes { and _
+        // Matches incomplete tokens starting with [, 【, ［, <, {, __, %5B, \u005b, or HTML entity &#91; / &lbrack;
         const tailLen = 160;
         const tail = this.pendingBuffer.length > tailLen ? this.pendingBuffer.slice(-tailLen) : this.pendingBuffer;
         const offset = this.pendingBuffer.length > tailLen ? this.pendingBuffer.length - tailLen : 0;
-        const m = tail.match(/(?:\[|<|\{\{?|__?)[A-Za-z0-9_\-А-Яа-яЁё ]{0,128}$/);
+        const m = tail.match(/(?:(?:\[|【|［|〔|「|『|<|\{\{?|__?|%5[Bb]|\\u005[Bb]|&(?:amp;)?(?:#0*91;|#[xX]0*5[Bb];|lbrack;))[A-Za-z0-9_\-А-Яа-яЁё *`~.\\]{0,128}(?:%5?|&(?:a(?:m(?:p(?:;)?)?)?)?(?:#(?:0*9?3?|[xX]0*5?[Dd]?)?|r(?:b(?:r(?:a(?:c(?:k)?)?)?)?)?)?)?|%[0-9A-Fa-f]?|\\(?:u(?:0(?:0(?:5)?)?)?)?|&(?:a(?:m(?:p(?:;)?)?)?)?(?:#(?:0*9?1?|[xX]0*5?)?|l(?:b(?:r(?:a(?:c(?:k)?)?)?)?)?)?)$/);
 
         if (m) {
             const candidate = m[0];
-            // Only closed double-underscore tokens (e.g. __NAME_1__) could match m since '_' is in character class.
-            // Bracketed tokens ([, <, {{) cannot end with their closing delimiter here and must be buffered.
-            const isClosedUnderscoreToken = candidate.length >= 4 && candidate.startsWith('__') && candidate.endsWith('__');
-            if (isClosedUnderscoreToken && this.replacer(candidate) !== candidate) {
+            // Closed double-underscore (__NAME_1__) or JSON Unicode (\u005bNAME_1\u005d) tokens match character class and can be emitted immediately.
+            const isClosedUnderscoreOrUnicode = (candidate.length >= 4 && candidate.startsWith('__') && candidate.endsWith('__')) || /\\u005[Dd]/.test(candidate);
+            if (isClosedUnderscoreOrUnicode && this.replacer(candidate) !== candidate) {
                 const toEmit = this.pendingBuffer;
                 this.pendingBuffer = '';
                 return this.replacer(toEmit);
@@ -773,6 +797,49 @@ export class StreamingDetokenizer {
         this.pendingBuffer = '';
         return this.replacer(remaining);
     }
+
+    /**
+     * Wraps an async iterable (e.g. LLM streaming response) yielding restored chunks.
+     * @param {AsyncIterable<any>} asyncIterable
+     * @returns {AsyncGenerator<any>}
+     */
+    async* wrapStream(asyncIterable) {
+        for await (const chunk of asyncIterable) {
+            if (chunk === null || chunk === undefined) continue;
+            if (chunk.choices && Array.isArray(chunk.choices) && chunk.choices[0]?.delta?.content !== undefined) {
+                const delta = chunk.choices[0].delta.content;
+                if (typeof delta === 'string') {
+                    const restored = this.feed(delta);
+                    if (restored) {
+                        chunk.choices[0].delta.content = restored;
+                        yield chunk;
+                    }
+                } else {
+                    yield chunk;
+                }
+                continue;
+            }
+
+            let text = '';
+            if (typeof chunk === 'string') {
+                text = chunk;
+            } else if (typeof chunk?.text === 'string') {
+                text = chunk.text;
+            } else if (typeof chunk?.content === 'string') {
+                text = chunk.content;
+            } else if (chunk instanceof Uint8Array || (typeof Buffer !== 'undefined' && Buffer.isBuffer(chunk))) {
+                text = chunk;
+            } else {
+                text = String(chunk);
+            }
+
+            const emitted = this.feed(text);
+            if (emitted) yield emitted;
+        }
+
+        const finalChunk = this.flush();
+        if (finalChunk) yield finalChunk;
+    }
 }
 
 /**
@@ -803,33 +870,198 @@ export async function* wrapAiStream(stream, tokenMap = {}, options = {}) {
 
 export function createStreamTokenRestorer(tokenMap = {}, options = {}) {
     const detok = new StreamingDetokenizer(tokenMap, options);
+    let stream;
     if (typeof TransformStream !== 'undefined') {
-        const ts = new TransformStream({
+        stream = new TransformStream({
             transform(chunk, controller) {
-                const str = typeof chunk === 'string' ? chunk : String(chunk);
-                const out = detok.process(str);
+                const out = detok.feed(chunk);
                 if (out) controller.enqueue(out);
             },
             flush(controller) {
-                const rem = detok.flush();
-                if (rem) controller.enqueue(rem);
+                const out = detok.flush();
+                if (out) controller.enqueue(out);
             }
         });
-        ts.process = (c) => detok.process(c);
-        ts.flush = () => detok.flush();
-        ts.detokenizer = detok;
-        return ts;
+    } else {
+        stream = {};
     }
-    return {
-        process: (c) => detok.process(c),
-        flush: () => detok.flush(),
-        detokenizer: detok
-    };
+    stream.process = (chunk) => detok.feed(chunk);
+    stream.feed = (chunk) => detok.feed(chunk);
+    stream.flush = () => detok.flush();
+    stream.wrapStream = (asyncIterable) => detok.wrapStream(asyncIterable);
+    stream.detokenizer = detok;
+    return stream;
 }
 
 export const scrubText = scrubberPkg ? scrubberPkg.scrubText : null;
 export const unscrubText = scrubberPkg ? scrubberPkg.unscrubText : null;
 
+// ── PSR-2 Receipt Integrity (real SHA-256; parity with src/modules/receipt-integrity.js) ──
+export const PSR2_VERSION = 'PSR-2';
+export const PSR2_STATEMENTS = {
+  measured: [
+    'Token Manifest SHA-256: digest of the sorted token labels held in this session (original values are never hashed or exported).',
+    'Entity counts per type.',
+    'Receipt SHA-256: digest of the canonical field string, so edits to this receipt are detectable.'
+  ],
+  byDesign: [
+    'The sanitization engine runs locally in process memory and does not transmit prompt content. This is an architectural property and is not measured by this receipt.',
+    'Token-to-original mappings are held in volatile memory for the session. This receipt does not record their contents.'
+  ],
+  limits: [
+    'This receipt is not a digital signature and does not prove who generated it or when.',
+    'No final output digest is recorded by this export (Output SHA-256: not recorded).',
+    'Detection is pattern-based. Free-text narrative can contain indirect identifiers that remain after sanitization; review output before sharing.',
+    'Regulatory references are informational technical-safeguard mappings, not a legal determination of compliance.'
+  ]
+};
+function psr2Sha256(input) {
+  const bytes = new TextEncoder().encode(String(input == null ? '' : input));
+  const K = [
+    0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5, 0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5,
+    0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3, 0x72be5d74, 0x80deb1fe, 0x9bdc06a7, 0xc19bf174,
+    0xe49b69c1, 0xefbe4786, 0x0fc19dc6, 0x240ca1cc, 0x2de92c6f, 0x4a7484aa, 0x5cb0a9dc, 0x76f988da,
+    0x983e5152, 0xa831c66d, 0xb00327c8, 0xbf597fc7, 0xc6e00bf3, 0xd5a79147, 0x06ca6351, 0x14292967,
+    0x27b70a85, 0x2e1b2138, 0x4d2c6dfc, 0x53380d13, 0x650a7354, 0x766a0abb, 0x81c2c92e, 0x92722c85,
+    0xa2bfe8a1, 0xa81a664b, 0xc24b8b70, 0xc76c51a3, 0xd192e819, 0xd6990624, 0xf40e3585, 0x106aa070,
+    0x19a4c116, 0x1e376c08, 0x2748774c, 0x34b0bcb5, 0x391c0cb3, 0x4ed8aa4a, 0x5b9cca4f, 0x682e6ff3,
+    0x748f82ee, 0x78a5636f, 0x84c87814, 0x8cc70208, 0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2
+  ];
+  const H = [0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a, 0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19];
+  const bitLen = bytes.length * 8;
+  const totalLen = (((bytes.length + 9 + 63) >> 6) << 6);
+  const msg = new Uint8Array(totalLen);
+  msg.set(bytes);
+  msg[bytes.length] = 0x80;
+  const view = new DataView(msg.buffer);
+  view.setUint32(totalLen - 8, Math.floor(bitLen / 0x100000000), false);
+  view.setUint32(totalLen - 4, bitLen >>> 0, false);
+  const W = new Uint32Array(64);
+  const rotr = (x, n) => (x >>> n) | (x << (32 - n));
+  for (let off = 0; off < totalLen; off += 64) {
+    for (let i = 0; i < 16; i++) W[i] = view.getUint32(off + i * 4, false);
+    for (let i = 16; i < 64; i++) {
+      const s0 = rotr(W[i - 15], 7) ^ rotr(W[i - 15], 18) ^ (W[i - 15] >>> 3);
+      const s1 = rotr(W[i - 2], 17) ^ rotr(W[i - 2], 19) ^ (W[i - 2] >>> 10);
+      W[i] = (W[i - 16] + s0 + W[i - 7] + s1) >>> 0;
+    }
+    let [a, b, c, d, e, f, g, h] = H;
+    for (let i = 0; i < 64; i++) {
+      const S1 = rotr(e, 6) ^ rotr(e, 11) ^ rotr(e, 25);
+      const ch = (e & f) ^ ((~e) & g);
+      const temp1 = (h + S1 + ch + K[i] + W[i]) >>> 0;
+      const S0 = rotr(a, 2) ^ rotr(a, 13) ^ rotr(a, 22);
+      const maj = (a & b) ^ (a & c) ^ (b & c);
+      const temp2 = (S0 + maj) >>> 0;
+      h = g; g = f; f = e; e = (d + temp1) >>> 0;
+      d = c; c = b; b = a; a = (temp1 + temp2) >>> 0;
+    }
+    H[0] = (H[0] + a) >>> 0; H[1] = (H[1] + b) >>> 0; H[2] = (H[2] + c) >>> 0; H[3] = (H[3] + d) >>> 0;
+    H[4] = (H[4] + e) >>> 0; H[5] = (H[5] + f) >>> 0; H[6] = (H[6] + g) >>> 0; H[7] = (H[7] + h) >>> 0;
+  }
+  return H.map(x => x.toString(16).padStart(8, '0')).join('');
+}
+export function buildPsr2Receipt(tokenMap, timestamp = new Date().toISOString()) {
+  const labels = (tokenMap && typeof tokenMap === 'object' ? Object.keys(tokenMap) : []).slice().sort();
+  const counts = {};
+  for (const t of labels) {
+    const m = t.match(/\[([A-Z_]+)_\d+\]/);
+    const type = m ? m[1] : 'CUSTOM';
+    counts[type] = (counts[type] || 0) + 1;
+  }
+  const entityCounts = Object.keys(counts).sort().map(k => `${k}=${counts[k]}`).join(',') || 'none';
+  const manifestSha256 = psr2Sha256(labels.join('\n'));
+  const canonical = [PSR2_VERSION, timestamp, 'none', manifestSha256, entityCounts, 'na'].join('|');
+  return { version: PSR2_VERSION, timestamp, outputSha256: null, manifestSha256, entityCounts, tokensTotal: labels.length, tokensPresentInOutput: null, canonical, receiptSha256: psr2Sha256(canonical) };
+}
+export function psr2Lines(r) {
+  return [
+    `Receipt Version: ${r.version}`,
+    `Timestamp (UTC): ${r.timestamp}`,
+    `Output SHA-256: not recorded`,
+    `Token Manifest SHA-256: ${r.manifestSha256}`,
+    `Entity Counts: ${r.entityCounts}`,
+    `Tokens In Final Output: not recorded`,
+    `Canonical: ${r.canonical}`,
+    `Receipt SHA-256: ${r.receiptSha256}`
+  ];
+}
+function formatPsr2Report({ format, receipt, telemetry, company_name, department, utmMedium }) {
+  const entitySummary = Object.entries(telemetry.entities).map(([t, count]) => `[${t}]: ${count}`).join(', ') || 'None (Clean)';
+  const teamsUrl = `https://privacyscrubber.com/teams?tier=teams&utm_source=sdk_audit_report&utm_medium=${utmMedium}&utm_campaign=teams_upgrade`;
+  const sdkUrl = `https://privacyscrubber.com/sdk/?tier=sdk&utm_source=sdk_audit_report&utm_medium=${utmMedium}&utm_campaign=sdk_pipeline`;
+  if (format === 'json') {
+    return JSON.stringify({
+      protocol: 'Zero-Trust Data Sanitization (ZTDS)',
+      receipt_version: receipt.version,
+      company: company_name,
+      department,
+      timestamp: receipt.timestamp,
+      output_sha256: null,
+      token_manifest_sha256: receipt.manifestSha256,
+      entity_counts: receipt.entityCounts,
+      canonical: receipt.canonical,
+      receipt_sha256: receipt.receiptSha256,
+      risk_level: telemetry.riskLevel,
+      frameworks_referenced: telemetry.frameworksList,
+      total_masked_tokens: telemetry.totalCount,
+      entities_breakdown: telemetry.entities,
+      scope: { measured: PSR2_STATEMENTS.measured, by_design_not_measured: PSR2_STATEMENTS.byDesign, limits: PSR2_STATEMENTS.limits },
+      verification_url: 'https://privacyscrubber.com/features/audit-receipt/#verify',
+      licensing_and_governance: { teams_governance_url: teamsUrl, sdk_pipeline_url: sdkUrl }
+    }, null, 2);
+  }
+  if (format === 'summary') {
+    return `[PrivacyScrubber SHA-256 Receipt] ${receipt.version} | Receipt SHA-256: ${receipt.receiptSha256.substring(0, 16)}... | Organization: ${company_name} | Masked Tokens: ${telemetry.totalCount} | Risk: ${telemetry.riskLevel} | Frameworks (informational): ${telemetry.frameworksList.join(', ')}`;
+  }
+  return `# Zero-Trust Data Sanitization Receipt
+**Organization:** ${company_name} (${department})  
+**Timestamp:** ${receipt.timestamp}  
+**Receipt Format:** ${receipt.version} (SHA-256 digests; not a digital signature)
+
+---
+
+### Sanitization Metrics & Risk Assessment
+* **Overall Risk Rating:** **${telemetry.riskLevel}**
+* **Total Sensitive Entities Masked:** \`${telemetry.totalCount}\`
+* **Entity Breakdown:** ${entitySummary}
+
+### Informational Framework References
+${telemetry.frameworksList.map(f => `- ${f}`).join('\n')}
+
+These are technical-safeguard mappings, not a legal determination of compliance.
+
+### Receipt Integrity (${receipt.version})
+\`\`\`text
+${psr2Lines(receipt).join('\n')}
+\`\`\`
+
+### Scope of This Receipt
+**Measured:** ${PSR2_STATEMENTS.measured.join(' ')}
+
+**Stated by design, not measured:** ${PSR2_STATEMENTS.byDesign.join(' ')}
+
+**Limits:** ${PSR2_STATEMENTS.limits.join(' ')}
+
+### Enterprise & Team Governance
+* **Team Governance & Encrypted Handoff:** Centrally deploy redaction rules and encrypted session handoff with **PrivacyScrubber TEAMS** ($99/mo flat): [privacyscrubber.com/teams](${teamsUrl})
+* **Backend Microservices & RAG Pipelines:** Integrate in-memory sanitization (<1ms) into your Node.js/Python infrastructure with **Developer SDK** ($299/mo): [privacyscrubber.com/sdk](${sdkUrl})
+
+*Generated locally by PrivacyScrubber Core SDK*  
+*Verify locally (paste this receipt): https://privacyscrubber.com/features/audit-receipt/#verify*`;
+}
+
+export function generateAuditReport(sessionMap = {}, options = {}) {
+    const { companyName = 'PrivacyScrubber Client', department = 'SecOps / Compliance', format = 'markdown' } = options;
+    const telemetry = buildAuditTelemetry(sessionMap);
+    const timestamp = new Date().toISOString();
+    const receipt = buildPsr2Receipt(sessionMap, timestamp);
+    if (format === 'json') {
+        return JSON.parse(formatPsr2Report({ format: 'json', receipt, telemetry, company_name: companyName, department, utmMedium: 'json' }));
+    }
+    return formatPsr2Report({ format: format === 'summary' ? 'summary' : 'markdown', receipt, telemetry, company_name: companyName, department, utmMedium: 'markdown' });
+}
+
 // Vision & Multimodal exports
 export {
     stripExifMetadata,
@@ -855,6 +1087,11 @@ export default {
     buildAuditTelemetry,
     buildCisoAuditTelemetry,
     formatAuditReceipt,
+    generateAuditReport,
+    PSR2_VERSION,
+    PSR2_STATEMENTS,
+    buildPsr2Receipt,
+    psr2Lines,
     PrivacyScrubberEngine,
     PrivacyScrubberLicenseError,
     StreamingDetokenizer,

package/cli.js

@@ -23,7 +23,7 @@ const __dirname = path.dirname(__filename);
 
 const args = process.argv.slice(2);
 
-let sdkVersion = '2.4.0';
+let sdkVersion = '2.4.1';
 try {
   const pkg = JSON.parse(fs.readFileSync(path.join(__dirname, 'package.json'), 'utf8'));
   sdkVersion = pkg.version || sdkVersion;

package/examples/python-bridge/bridge_daemon.cjs

@@ -24,6 +24,7 @@ const candidatePaths = [
     process.env.PRIVACYSCRUBBER_SDK_PATH,
     path.resolve(process.cwd(), 'npm-sdk/index.cjs'),
     path.resolve(process.cwd(), 'node_modules/@privacyscrubber/sdk'),
+    path.resolve(__dirname, '../../../index.cjs'),
     path.resolve(__dirname, '../../index.cjs'),
     path.resolve(__dirname, '../index.cjs'),
     path.resolve(__dirname, './index.cjs'),
@@ -115,7 +116,7 @@ rl.on('line', (line) => {
                 sendResponse(id, {
                     pong: true,
                     engine: 'PrivacyScrubber ZTDS',
-                    version: '2.4.0',
+                    version: '2.4.1',
                     uptime: process.uptime()
                 });
                 break;
@@ -171,7 +172,6 @@ rl.on('line', (line) => {
                 };
 
                 const res = sdk.sanitizeObject(data, options);
-                // Also merge into engine sessionMap if requested
                 if (req.persist_session !== false && res.tokenMap) {
                     Object.assign(engine.sessionMap, res.tokenMap);
                 }

package/examples/python-bridge/fastapi_example.py

@@ -59,7 +59,7 @@ if FASTAPI_AVAILABLE:
     app = FastAPI(
         title="PrivacyScrubber ZTDS Sidecar Proxy",
         description="Zero-Trust Data Sanitization & In-Memory PII Shield for LLM Pipelines",
-        version="2.4.0"
+        version="2.4.1"
     )
 
     @app.on_event("startup")

package/examples/python-bridge/privacyscrubber/__init__.py

@@ -24,7 +24,7 @@ from privacyscrubber.exceptions import (
     PrivacyScrubberLicenseError,
 )
 
-__version__ = "2.4.0"
+__version__ = "2.4.1"
 __author__ = "Ilya Sibiryakov (BrandMeWeb)"
 __license__ = "Proprietary"
 

package/examples/python-bridge/privacyscrubber/bridge_daemon.cjs

@@ -116,7 +116,7 @@ rl.on('line', (line) => {
                 sendResponse(id, {
                     pong: true,
                     engine: 'PrivacyScrubber ZTDS',
-                    version: '2.4.0',
+                    version: '2.4.1',
                     uptime: process.uptime()
                 });
                 break;

package/examples/python-bridge/privacyscrubber/cli.py

@@ -50,7 +50,7 @@ def main() -> None:
     args = parser.parse_args()
 
     if args.version:
-        print("PrivacyScrubber ZTDS Python SDK v2.4.0")
+        print("PrivacyScrubber ZTDS Python SDK v2.4.1")
         sys.exit(0)
 
     if not args.command:

package/examples/python-bridge/privacyscrubber/client.py

@@ -89,14 +89,16 @@ class PrivacyScrubberClient:
                 self.daemon_path,
                 "--profile", self.default_profile
             ]
+            proc_env = os.environ.copy()
             if self.license_key:
-                cmd.extend(["--license-key", self.license_key])
+                proc_env["PRIVACYSCRUBBER_LICENSE_KEY"] = self.license_key
 
             self._process = subprocess.Popen(
                 cmd,
                 stdin=subprocess.PIPE,
                 stdout=subprocess.PIPE,
                 stderr=subprocess.PIPE,
+                env=proc_env,
                 text=True,
                 bufsize=1
             )
@@ -341,14 +343,16 @@ class AsyncPrivacyScrubberClient:
                 self.daemon_path,
                 "--profile", self.default_profile
             ]
+            proc_env = os.environ.copy()
             if self.license_key:
-                cmd.extend(["--license-key", self.license_key])
+                proc_env["PRIVACYSCRUBBER_LICENSE_KEY"] = self.license_key
 
             self._process = await asyncio.create_subprocess_exec(
                 *cmd,
                 stdin=asyncio.subprocess.PIPE,
                 stdout=asyncio.subprocess.PIPE,
-                stderr=asyncio.subprocess.PIPE
+                stderr=asyncio.subprocess.PIPE,
+                env=proc_env
             )
 
     async def close(self) -> None:
@@ -387,13 +391,15 @@ class AsyncPrivacyScrubberClient:
                     self.daemon_path,
                     "--profile", self.default_profile
                 ]
+                proc_env = os.environ.copy()
                 if self.license_key:
-                    cmd.extend(["--license-key", self.license_key])
+                    proc_env["PRIVACYSCRUBBER_LICENSE_KEY"] = self.license_key
                 self._process = await asyncio.create_subprocess_exec(
                     *cmd,
                     stdin=asyncio.subprocess.PIPE,
                     stdout=asyncio.subprocess.PIPE,
-                    stderr=asyncio.subprocess.PIPE
+                    stderr=asyncio.subprocess.PIPE,
+                    env=proc_env
                 )
 
             self._req_id += 1