@privacyscrubber/sdk 2.3.7 → 2.4.0

package/index.cjs

@@ -36,16 +36,33 @@ if (scrubberPkg && typeof scrubberPkg.init === 'function') {
  * Built-in patterns for DevOps & Cloud Secrets detection.
  */
 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' }
 ];
@@ -141,8 +158,8 @@ function buildAuditTelemetry(currentTokenMap = {}) {
         frameworksSet.add('CCPA/CPRA');
     }
     if (types.includes('ID') || types.includes('SSN') || types.includes('PASSPORT') || types.includes('NATIONAL_ID')) {
-        frameworksSet.add('SOC 2 Type II');
-        frameworksSet.add('ISO 27001 (A.8.11)');
+        frameworksSet.add('SOC 2 controls');
+        frameworksSet.add('ISO 27001 A.8.11 controls');
     }
     if (types.some(t => ['CREDIT_CARD', 'BANK', 'IBAN', 'FINANCIAL', 'CARD', 'STRIPE_KEY', 'TAX_ID', 'EIN', 'W2', '1099', 'SALARY', 'WAGE'].includes(t))) {
         frameworksSet.add('PCI DSS v4.0');
@@ -231,7 +248,7 @@ function formatAuditReceipt(telemetry) {
 
     return `> 🛡️ **PrivacyScrubber Audit Receipt**\n` +
            `> * **Risk Level:** ${icon} ${telemetry.riskLevel}\n` +
-           `> * **Compliance Enforced:** ${telemetry.frameworksList.join(', ')}\n` +
+           `> * **Framework References (informational):** ${telemetry.frameworksList.join(', ')}\n` +
            `> * **Tokens Masked:** ${telemetry.totalCount} (${entitiesList})\n` +
            `> * ⭐ **Star on GitHub:** [moxno/privacyscrubber-mcp](https://github.com/moxno/privacyscrubber-mcp) | **SDK & Enterprise:** [privacyscrubber.com/pricing](https://privacyscrubber.com/pricing)\n`;
 }
@@ -1466,13 +1483,13 @@ 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);
@@ -1501,10 +1518,14 @@ 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);
                     }
@@ -1610,18 +1631,17 @@ 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);
@@ -1843,13 +1863,8 @@ function withPrivacyScrubber(handler, options = {}) {
             tokenMap = res.tokenMap;
             telemetry = res.telemetry;
         } else if (rawBody && typeof rawBody === 'object') {
-            const serialized = JSON.stringify(rawBody);
-            const res = sanitize(serialized, options);
-            try {
-                scrubbedBody = JSON.parse(res.scrubbedText);
-            } catch (_) {
-                scrubbedBody = rawBody;
-            }
+            const res = sanitizeObject(rawBody, options);
+            scrubbedBody = res.sanitized;
             tokenMap = res.tokenMap;
             telemetry = res.telemetry;
         }
@@ -1860,7 +1875,8 @@ function withPrivacyScrubber(handler, options = {}) {
             scrubbedBody,
             tokenMap,
             telemetry,
-            restore: (text) => restore(text, tokenMap),
+            restore: (text) => (text && typeof text === 'object') ? restoreObject(text, tokenMap) : restore(text, tokenMap),
+            restoreObject: (obj) => restoreObject(obj, tokenMap),
             createStreamRestorer: () => createStreamTokenRestorer(tokenMap)
         };
 
@@ -1878,15 +1894,13 @@ function withPrivacyScrubber(handler, options = {}) {
 function createExpressMiddleware(options = {}) {
     return function privacyScrubberExpress(req, res, next) {
         if (req && req.body && typeof req.body === 'object') {
-            const serialized = JSON.stringify(req.body);
-            const { scrubbedText, tokenMap, telemetry } = sanitize(serialized, options);
+            const { sanitized, tokenMap, telemetry } = sanitizeObject(req.body, options);
             req.rawBody = req.body;
-            try {
-                req.body = JSON.parse(scrubbedText);
-            } catch (_) {}
+            req.body = sanitized;
             req.psTokenMap = tokenMap;
             req.psTelemetry = telemetry;
-            req.restore = (aiResponse) => restore(aiResponse, tokenMap);
+            req.restore = (aiResponse) => (aiResponse && typeof aiResponse === 'object') ? restoreObject(aiResponse, tokenMap) : restore(aiResponse, tokenMap);
+            req.restoreObject = (obj) => restoreObject(obj, tokenMap);
             req.createStreamRestorer = () => createStreamTokenRestorer(tokenMap);
         }
         if (typeof next === 'function') next();
@@ -1903,7 +1917,7 @@ function formatCefEvent(eventData = {}) {
     const {
         vendor = 'PrivacyScrubber',
         product = 'ZTDS',
-        version = '2.2.5',
+        version = '2.4.0',
         eventClassId = 'PII_REDACTION',
         eventName = 'Zero-Trust PII Sanitization Executed',
         riskLevel = 'LOW EXPOSURE',
@@ -1972,7 +1986,7 @@ function formatSyslogEvent(eventData = {}) {
 function formatJsonlEvent(eventData = {}) {
     const {
         timestamp = new Date().toISOString(),
-        version = '2.2.5',
+        version = '2.4.0',
         protocol = 'ZTDS',
         event = 'sanitization',
         riskLevel = 'LOW EXPOSURE',
@@ -1997,7 +2011,7 @@ function formatJsonlEvent(eventData = {}) {
         frameworks,
         session_hash: sessionHash,
         zero_egress: true,
-        verification_mode: "100% Offline (In-Memory RAM)",
+        processing_mode: "local in-memory (by design, not measured)",
         ...(organization && { organization }),
         ...(department && { department }),
         ...(executionMs && { execution_ms: executionMs }),
@@ -2089,13 +2103,7 @@ function createAuditLogger(options = {}) {
             telemetry = buildAuditTelemetry(tokenMap);
         }
 
-        let sessionHash = '';
-        try {
-            const cryptoMod = require('crypto');
-            sessionHash = cryptoMod.createHash('sha256').update(JSON.stringify(tokenMap) + Date.now()).digest('hex');
-        } catch {
-            sessionHash = 'ztds_' + Math.random().toString(16).substring(2, 10);
-        }
+        const sessionHash = buildPsr2Receipt(tokenMap).receiptSha256;
 
         const eventData = {
             event: 'pii_sanitization',
@@ -2141,108 +2149,138 @@ function createAuditLogger(options = {}) {
     };
 }
 
+// ── PSR-2 Receipt Integrity (real SHA-256; parity with src/modules/receipt-integrity.js) ──
+const PSR2_VERSION = 'PSR-2';
+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) {
+  return require('crypto').createHash('sha256').update(String(input == null ? '' : input), 'utf8').digest('hex');
+}
+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) };
+}
+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*`;
+}
+
 function generateAuditReport(sessionMap = {}, options = {}) {
     const { companyName = 'PrivacyScrubber Client', department = 'SecOps / Compliance', format = 'markdown' } = options;
     const telemetry = buildAuditTelemetry(sessionMap);
-    
-    let sessionHash = '';
-    try {
-        const cryptoMod = require('crypto');
-        sessionHash = cryptoMod.createHash('sha256').update(JSON.stringify(sessionMap) + Date.now().toString()).digest('hex');
-    } catch {
-        sessionHash = Array.from(crypto.getRandomValues(new Uint8Array(16))).map(b => b.toString(16).padStart(2, '0')).join('');
-    }
-
     const timestamp = new Date().toISOString();
-    const entitySummary = Object.entries(telemetry.entities).map(([t, count]) => `[${t}]: ${count}`).join(', ') || 'None (Clean)';
+    const receipt = buildPsr2Receipt(sessionMap, timestamp);
+    const sessionHash = receipt.receiptSha256;
 
     if (format === 'cef') {
-        return formatCefEvent({
-            organization: companyName,
-            department: department,
-            timestamp,
-            sessionHash,
-            riskLevel: telemetry.riskLevel,
-            tokensMasked: telemetry.totalCount,
-            entities: telemetry.entities,
-            frameworks: telemetry.frameworksList
-        });
+        return formatCefEvent({ organization: companyName, department, timestamp, sessionHash, riskLevel: telemetry.riskLevel, tokensMasked: telemetry.totalCount, entities: telemetry.entities, frameworks: telemetry.frameworksList });
     }
-
     if (format === 'syslog') {
-        return formatSyslogEvent({
-            organization: companyName,
-            department: department,
-            timestamp,
-            sessionHash,
-            riskLevel: telemetry.riskLevel,
-            tokensMasked: telemetry.totalCount,
-            entities: telemetry.entities,
-            frameworks: telemetry.frameworksList
-        });
+        return formatSyslogEvent({ organization: companyName, department, timestamp, sessionHash, riskLevel: telemetry.riskLevel, tokensMasked: telemetry.totalCount, entities: telemetry.entities, frameworks: telemetry.frameworksList });
     }
-
     if (format === 'jsonl') {
-        return formatJsonlEvent({
-            organization: companyName,
-            department: department,
-            timestamp,
-            sessionHash,
-            riskLevel: telemetry.riskLevel,
-            tokensMasked: telemetry.totalCount,
-            entities: telemetry.entities,
-            frameworks: telemetry.frameworksList
-        });
+        return formatJsonlEvent({ organization: companyName, department, timestamp, sessionHash, riskLevel: telemetry.riskLevel, tokensMasked: telemetry.totalCount, entities: telemetry.entities, frameworks: telemetry.frameworksList });
     }
-
     if (format === 'json') {
-        return {
-            protocol: "Zero-Trust Data Sanitization (ZTDS)",
-            certificateId: `ZTDS-CERT-${sessionHash.substring(0, 16).toUpperCase()}`,
-            organization: companyName,
-            department: department,
-            timestamp: timestamp,
-            sessionHash: sessionHash,
-            verificationMode: "100% Offline (In-Memory RAM)",
-            complianceStatus: "VERIFIED PASS",
-            riskLevel: telemetry.riskLevel,
-            frameworksEnforced: telemetry.frameworksList,
-            totalMaskedTokens: telemetry.totalCount,
-            entitiesBreakdown: telemetry.entities,
-            zeroEgressVerified: true,
-            verificationUrl: `https://privacyscrubber.com/features/audit-receipt/#verify?hash=${sessionHash.substring(0, 16)}`
-        };
+        return JSON.parse(formatPsr2Report({ format: 'json', receipt, telemetry, company_name: companyName, department, utmMedium: 'json' }));
     }
-
-    if (format === 'summary') {
-        return `[PrivacyScrubber Compliance Certificate] ID: ZTDS-${sessionHash.substring(0, 8).toUpperCase()} | Organization: ${companyName} | Masked Tokens: ${telemetry.totalCount} | Risk: ${telemetry.riskLevel} | Frameworks: ${telemetry.frameworksList.join(', ')} | Status: PASS (100% Air-Gapped)`;
-    }
-
-    return `# 🛡️ Zero-Trust Data Sanitization Compliance Certificate
-**Certificate ID:** \`ZTDS-CERT-${sessionHash.substring(0, 16).toUpperCase()}\`  
-**Organization:** ${companyName} (${department})  
-**Timestamp:** ${timestamp}  
-**Verification Mode:** 100% Local In-Memory Processing (Air-Gapped)  
-**Status:** **VERIFIED PASS** (Zero Network Egress)
-
----
-
-### 📊 Sanitization Metrics & Risk Assessment
-* **Overall Risk Rating:** **${telemetry.riskLevel}**
-* **Total Sensitive Entities Masked:** \`${telemetry.totalCount}\`
-* **Entity Breakdown:** ${entitySummary}
-* **Network Data Transmitted:** \`0.00 KB (Zero-Trust Local RAM)\`
-
-### 📜 Regulatory Frameworks Enforced
-${telemetry.frameworksList.map(f => `- **${f}**`).join('\n')}
-
-### 🔒 CISO Compliance Declaration
-1. **EU AI Act (Art. 50) & GDPR (Art. 25 & 32):** Data minimization and local pseudonymization enforced prior to model interaction.
-2. **HIPAA Safe Harbor (§164.514) / SOC 2 Type II:** All direct and indirect identifiers sanitized locally without cloud processor liability.
-3. **Cryptographic Verification:** Tamper-evident session verification hash: \`${sessionHash}\`
-
-*Certified Offline by PrivacyScrubber Core SDK*  
-*Verify at: https://privacyscrubber.com/features/audit-receipt/*`;
+    return formatPsr2Report({ format: format === 'summary' ? 'summary' : 'markdown', receipt, telemetry, company_name: companyName, department, utmMedium: 'markdown' });
 }
 
 /**
@@ -2484,7 +2522,46 @@ function createGuardedTools(engineOrOptions = {}, toolConfig = {}) {
                     throw new Error(`[guard_read_file] Path is a directory, not a file: ${resolved}`);
                 }
 
-                const raw = fs.readFileSync(resolved, 'utf8');
+                const rawBuf = fs.readFileSync(resolved);
+                const checkLen = Math.min(rawBuf.length, 8000);
+                for (let i = 0; i < checkLen; i++) {
+                    if (rawBuf[i] === 0) {
+                        throw new Error(`[guard_read_file] Binary file detected (${path.basename(resolved)}). Only text, code, and MIME files are supported.`);
+                    }
+                }
+                let raw = rawBuf.toString('utf8');
+                const ext = path.extname(resolved).toLowerCase();
+                if (ext === '.eml') {
+                    const decodeQpUtf8 = (str) =>
+                        str.replace(/=\r?\n/g, '').replace(/(?:=[0-9A-Fa-f]{2})+/g, (seq) => {
+                            try {
+                                const bytes = Buffer.alloc(seq.length / 3);
+                                for (let i = 0; i < bytes.length; i++) {
+                                    bytes[i] = parseInt(seq.slice(i * 3 + 1, i * 3 + 3), 16);
+                                }
+                                return bytes.toString('utf8');
+                            } catch (_) { return seq; }
+                        });
+                    raw = raw.replace(/=\?([^?]+)\?([BQbq])\?([^?]+)\?=/g, (full, _cs, enc, payload) => {
+                        try {
+                            return enc.toUpperCase() === 'B'
+                                ? Buffer.from(payload.replace(/\s+/g, ''), 'base64').toString('utf8')
+                                : decodeQpUtf8(payload.replace(/_/g, ' '));
+                        } catch (_) { return full; }
+                    });
+                    raw = decodeQpUtf8(raw);
+                } else if (ext === '.rtf') {
+                    raw = raw
+                        .replace(/\\par\b|\\line\b/gi, '\n')
+                        .replace(/\\'([0-9a-fA-F]{2})/g, (_, hex) => String.fromCharCode(parseInt(hex, 16)))
+                        .replace(/\\u(-?\d+)\??/g, (_, code) => {
+                            const n = parseInt(code, 10);
+                            return String.fromCharCode(n < 0 ? n + 65536 : n);
+                        })
+                        .replace(/\\[a-zA-Z]+(?:-?\d+)?[ ]?/g, ' ')
+                        .replace(/[{}]/g, ' ')
+                        .replace(/[ \t]+/g, ' ');
+                }
                 const lines = raw.split(/\r?\n/);
                 const capped = lines.slice(0, maxLines).join('\n');
                 const wasTruncated = lines.length > maxLines;