@astrasyncai/verification-gateway 5.8.0 → 5.10.0

package/dist/edge-core/index.mjs

@@ -1,5 +1,5 @@
 // src/version.ts
-var SDK_VERSION = "5.8.0";
+var SDK_VERSION = "5.10.0";
 
 // src/http.ts
 var SDK_USER_AGENT = `astrasync-sdk/${SDK_VERSION}`;
@@ -25,7 +25,7 @@ var DEFAULT_EDGE_CONFIG = Object.freeze({
   mode: "observe",
   depth: "classify",
   pathRules: [],
-  sampling: Object.freeze({ anonymousBeaconRate: 0.1 }),
+  sampling: Object.freeze({ anonymousBeaconRate: 1 }),
   failurePosture: "open"
 });
 var DEFAULT_TTL_MS = 60 * 1e3;
@@ -111,6 +111,12 @@ var KEY_FORMAT_PREFIXES = [
 var MAX_HEADERS = 64;
 var MAX_HEADER_VALUE_BYTES = 1024;
 var MAX_TOTAL_BYTES = 16 * 1024;
+var DECLARED_SOURCE_MAX = 64;
+var DECLARED_SOURCE_PATTERN = /^[a-z0-9][a-z0-9 ._/-]{0,63}$/;
+function normalizeDeclaredSource(value) {
+  const trimmed = value?.trim().slice(0, DECLARED_SOURCE_MAX).toLowerCase();
+  return trimmed && DECLARED_SOURCE_PATTERN.test(trimmed) ? trimmed : void 0;
+}
 function byteLength(s) {
   let bytes = 0;
   for (let i = 0; i < s.length; i++) {
@@ -214,7 +220,20 @@ var PLATFORM_AGENT_SIGNATURES = [
   {
     vendor: "gemini",
     displayName: "Google Gemini",
-    uaPatterns: [/gemini/i, /google-gemini/i, /bard/i],
+    uaPatterns: [
+      /gemini/i,
+      /google-gemini/i,
+      /bard/i,
+      // Google's AI/agent fetcher tokens (VI uplift 1.3). These are the
+      // AI-training/agent crawlers, NOT the Search crawler — googlebot IP
+      // ranges attribute Search and are deliberately absent from
+      // vendor-ip-ranges. Anchored tokens: none appears in stock browser
+      // UAs, and `-` counts as a \b boundary so GoogleOther-Image/-Video
+      // match too.
+      /\bGoogle-Extended\b/i,
+      /\bGoogleOther\b/i,
+      /\bGoogle-CloudVertexBot\b/i
+    ],
     agentCardPatterns: [/gemini\.google\.com/i, /ai\.google\.dev/i],
     apiKeyFormatPatterns: [/^aiza/i],
     headerPatterns: [/^x-goog-user-project/i, /^x-goog-api-client/i]
@@ -260,6 +279,16 @@ var PLATFORM_AGENT_SIGNATURES = [
     uaPatterns: [/\bcomet\/\d/i],
     headerPatterns: [/^sec-ch-ua: .*comet/i]
   },
+  // The MCP bridge's own catalog fetches (VI uplift 1.5). Deliberately a
+  // DISTINCT UA from `astrasync-sdk/`: bridge fetches are buyer-agent
+  // traffic merchants should see (identified · unregistered), so this
+  // vendor is kept OUT of the self-traffic filter and the edge's SELF
+  // short-circuit.
+  {
+    vendor: "astrasync-bridge",
+    displayName: "AstraSync MCP bridge (buyer-agent catalog fetch)",
+    uaPatterns: [/^astrasync-bridge\//i]
+  },
   // Our own SDK self-identifies (sdkFetch stamps this UA) — first-party
   // traffic must never pollute the unknown/anonymous pool it exists to audit.
   {
@@ -651,6 +680,24 @@ async function callVerifyAccessAPI(config, request) {
         }
       };
     }
+    if (response.status === 400 && data?.error === "INVALID_AGENT_ID") {
+      const message = typeof data?.message === "string" ? data.message : "Agent identifier must be a canonical ASTRA-* id.";
+      return {
+        success: true,
+        attemptId: typeof data?.attemptId === "string" ? data.attemptId : void 0,
+        access: {
+          allowed: false,
+          reason: "invalid_agent_id",
+          failures: [
+            {
+              dimension: "identity.invalid_agent_id",
+              message,
+              guidance: typeof data?.guidance === "string" ? data.guidance : "Present your canonical ASTRA-* id (discover it via GET /api/agents/request-registration/{requestId} after owner approval)."
+            }
+          ]
+        }
+      };
+    }
     if (!response.ok) {
       if (response.status === 401) {
         return {
@@ -790,7 +837,7 @@ async function verify(config, request, options) {
     name: apiResponse.agent.name,
     trustScore: apiResponse.agent.trustScore,
     trustLevel: getTrustLevel(apiResponse.agent.trustScore),
-    blockchainVerified: apiResponse.agent.blockchainStatus === "verified",
+    blockchainAnchored: apiResponse.agent.blockchainStatus === "anchored",
     status: apiResponse.agent.agentStatus
   } : void 0;
   const developer = apiResponse.developer ? {
@@ -890,10 +937,24 @@ function requestUrl(req, host) {
   const qs = req.querystring ? `?${req.querystring}` : "";
   return `https://${host}${req.uri}${qs}`;
 }
-function buildObservedMetadata(req) {
+var ATTEMPT_ID_RE = /^att_[0-9a-f]{32}$/;
+function extractAttemptId(req) {
+  const header = headerValue(req, "x-astra-attempt-id");
+  if (header && ATTEMPT_ID_RE.test(header)) return header;
+  if (req.querystring) {
+    try {
+      const fromQuery = new URLSearchParams(req.querystring).get("astra_attempt");
+      if (fromQuery && ATTEMPT_ID_RE.test(fromQuery)) return fromQuery;
+    } catch {
+    }
+  }
+  return void 0;
+}
+function buildObservedMetadata(req, queryValueAllowlist) {
   const { headers, apiKeyFormat, platformHeaders, cookieCount } = sanitizeHeaders(req.headers);
   const connection = req.connection ?? {};
   const queryKeys = extractQueryKeys(req.querystring);
+  const queryParams = extractAllowlistedQueryParams(req.querystring, queryValueAllowlist);
   return {
     headers,
     ...apiKeyFormat && { apiKeyFormat },
@@ -901,6 +962,7 @@ function buildObservedMetadata(req) {
     ...Object.keys(connection).length > 0 && { connection },
     ...cookieCount !== void 0 && { cookieCount },
     ...queryKeys && { queryKeys },
+    ...queryParams && { queryParams },
     schemaVersion: CAPTURE_SCHEMA_VERSION
   };
 }
@@ -924,6 +986,29 @@ function extractQueryKeys(querystring) {
   }
   return keys.length > 0 ? keys : void 0;
 }
+var MAX_QUERY_PARAMS = 16;
+var MAX_QUERY_VALUE_CHARS = 256;
+function extractAllowlistedQueryParams(querystring, allowlist) {
+  if (!querystring || !allowlist || allowlist.length === 0) return void 0;
+  let params;
+  try {
+    params = new URLSearchParams(querystring);
+  } catch {
+    return void 0;
+  }
+  const matches = (name) => allowlist.some(
+    (entry) => entry.endsWith("*") ? name.startsWith(entry.slice(0, -1)) : name === entry
+  );
+  const out = {};
+  let count = 0;
+  for (const [key, value] of params.entries()) {
+    const name = key.toLowerCase().slice(0, MAX_QUERY_KEY_CHARS);
+    if (!name || out[name] !== void 0 || !matches(name)) continue;
+    out[name] = value.slice(0, MAX_QUERY_VALUE_CHARS);
+    if (++count >= MAX_QUERY_PARAMS) break;
+  }
+  return count > 0 ? out : void 0;
+}
 
 // src/edge-core/detect.ts
 var SD_JWT_PATTERN = /^eyJ[\w-]+\.[\w-]*\.[\w-]+~/;
@@ -1083,7 +1168,8 @@ function resolvePolicy(config, path) {
   return {
     mode: rule.mode ?? config.mode,
     depth: rule.depth ?? config.depth,
-    skip: rule.skip === true
+    skip: rule.skip === true,
+    ...rule.sampling && { sampling: rule.sampling }
   };
 }
 function classify(req, host, depth) {
@@ -1095,7 +1181,7 @@ function classify(req, host, depth) {
   if (!credentials.astraId && (isAccountApiKey(credentials.jwt) || isAccountApiKey(credentials.apiKey))) {
     return { tier: "self", evidence: { firstPartyApiKey: true } };
   }
-  if (credentials.astraId || credentials.apiKey || credentials.jwt) {
+  if (credentials.astraId) {
     return { tier: "verified", protocol: "x-astra" };
   }
   const { apiKeyFormat, platformHeaders } = sanitizeHeaders(req.headers);
@@ -1377,7 +1463,9 @@ async function withBudget(promise, ms) {
     if (timer) clearTimeout(timer);
   }
 }
-async function emitBeacon(options, req, classification, budgetMs) {
+async function emitBeacon(options, req, classification, budgetMs, queryValueAllowlist) {
+  const attemptId = extractAttemptId(req);
+  const declaredSource = normalizeDeclaredSource(headerValue(req, "x-astra-source"));
   const budgeted = withBudget(
     reportUnregisteredAttempt(
       { apiBaseUrl: options.apiBaseUrl, counterpartyId: options.counterpartyId },
@@ -1388,12 +1476,20 @@ async function emitBeacon(options, req, classification, budgetMs) {
         userAgent: headerValue(req, "user-agent"),
         requestPath: req.uri,
         requestMethod: req.method,
+        // Attempt-chain handle (header or ?astra_attempt= deep link) — joins
+        // this beacon into the agent's research chain via pe_attempt_idx.
+        ...attemptId && { attemptId },
+        // First-class referer + accept: traffic-source and surface-classifier
+        // inputs (previously buried inside observedMetadata.headers).
+        referer: headerValue(req, "referer"),
+        accept: headerValue(req, "accept"),
+        ...declaredSource && { declaredSource },
         // Evidence fields the backend persists into the event record.
         protocol: classification.protocol,
         evidence: classification.evidence,
         // The full sanitized header + connection capture for the
         // unregistered caller (fire-and-forget, so free).
-        observedMetadata: buildObservedMetadata(req)
+        observedMetadata: buildObservedMetadata(req, queryValueAllowlist)
       }
     ),
     budgetMs
@@ -1419,13 +1515,14 @@ function baseOutcome(classification) {
     protocol: classification.protocol
   };
 }
-function buildVerificationRequest(options, req, classification, includeArtifacts) {
+function buildVerificationRequest(options, req, classification, includeArtifacts, queryValueAllowlist) {
   const pdlss = resolveHttpPdlss({
     method: req.method,
     astraPurpose: headerValue(req, "x-astra-purpose"),
     astraAction: headerValue(req, "x-astra-action")
   });
   const attemptId = headerValue(req, "x-astra-attempt-id");
+  const declaredSource = normalizeDeclaredSource(headerValue(req, "x-astra-source"));
   return {
     ...attemptId && /^att_[0-9a-f]{32}$/.test(attemptId) && { attemptId },
     credentials: extractCredentials(req.headers, {}),
@@ -1440,10 +1537,11 @@ function buildVerificationRequest(options, req, classification, includeArtifacts
       userAgent: headerValue(req, "user-agent"),
       referer: headerValue(req, "referer"),
       host: headerValue(req, "host"),
+      ...declaredSource && { declaredSource },
       // The full sanitized header + connection capture. The backend stores
       // it fire-and-forget (zero decision latency) and uses the key-format
       // / platform headers for attribution.
-      observedMetadata: buildObservedMetadata(req)
+      observedMetadata: buildObservedMetadata(req, queryValueAllowlist)
     },
     ...includeArtifacts && classification.artifacts && { commerceArtifacts: classification.artifacts }
   };
@@ -1490,7 +1588,8 @@ function unavailableOutcome(classification, policy, reason, correlationId) {
 function isApiErrorResult(result) {
   return (result.failures ?? []).some((f) => f.dimension === "verify_access.api_error") === true && !result.misconfigured;
 }
-async function runTier(options, req, classification, policy, apiKey, sampling) {
+async function runTier(options, req, classification, policy, apiKey, capture) {
+  const { sampling, queryValueAllowlist } = capture;
   const budgets = options.budgets ?? {};
   const beaconMs = budgets.beaconMs ?? 300;
   const verifyMs = policy.mode === "enforce" ? budgets.verifyEnforceMs ?? 3e3 : budgets.verifyObserveMs ?? 1500;
@@ -1511,7 +1610,13 @@ async function runTier(options, req, classification, policy, apiKey, sampling) {
         const result = await withBudget(
           verify(
             gatewayConfig,
-            buildVerificationRequest(options, req, classification, policy.depth === "authorize")
+            buildVerificationRequest(
+              options,
+              req,
+              classification,
+              policy.depth === "authorize",
+              queryValueAllowlist
+            )
           ),
           verifyMs
         );
@@ -1535,7 +1640,10 @@ async function runTier(options, req, classification, policy, apiKey, sampling) {
       case "identified": {
         if (policy.depth === "authorize" && classification.artifacts) {
           const result = await withBudget(
-            verify(gatewayConfig, buildVerificationRequest(options, req, classification, true)),
+            verify(
+              gatewayConfig,
+              buildVerificationRequest(options, req, classification, true, queryValueAllowlist)
+            ),
             verifyMs
           );
           if (!result) {
@@ -1555,7 +1663,7 @@ async function runTier(options, req, classification, policy, apiKey, sampling) {
           }
           return outcomeFromVerifyResult(classification, result);
         }
-        await emitBeacon(options, req, classification, beaconMs);
+        await emitBeacon(options, req, classification, beaconMs, queryValueAllowlist);
         const outcome = baseOutcome(classification);
         if (policy.mode === "enforce" && policy.depth !== "classify") {
           outcome.decision = "deny";
@@ -1569,7 +1677,8 @@ async function runTier(options, req, classification, policy, apiKey, sampling) {
             options,
             req,
             { ...classification, protocol: classification.protocol ?? "ua-heuristic" },
-            beaconMs
+            beaconMs,
+            queryValueAllowlist
           );
         }
         return baseOutcome(classification);
@@ -1596,7 +1705,10 @@ async function evaluateEdgeRequest(options, req, config, apiKey) {
     if (classification.tier === "human" || classification.tier === "self") {
       return { policy, classification, application: passUntouched() };
     }
-    const outcome = await runTier(options, req, classification, policy, apiKey, config.sampling);
+    const outcome = await runTier(options, req, classification, policy, apiKey, {
+      sampling: policy.sampling ?? config.sampling,
+      queryValueAllowlist: config.queryValueAllowlist
+    });
     if (outcome.decision !== "allow") {
       options.log?.(
         "info",