@skrr-ai/cli 0.1.103 → 0.1.105

package/dist/node_modules/@skrr-ai/auth-core/dist/esm/telemetryRedaction.js

@@ -0,0 +1,216 @@
+/**
+ * Keeping a Hosted Machine preview capability out of telemetry (OSK-16174).
+ *
+ * `/api/previews/<token>/...` serves a person's machine with NO login: the token in the path is a bearer
+ * capability (`pvw_<22 chars>.<expiry>.<22 chars>`, idle TTL ten minutes). Anyone who can read a trace, an error
+ * report or a log line that carries it can open the preview, so every sink that records a URL has to run its
+ * strings through here first. This file is the ONE place that knows what to look for; the sinks (OpenTelemetry
+ * exporters, Sentry hooks, request and CSRF logs, client-log ingest) only decide WHERE to call it.
+ *
+ * WHAT IT MATCHES. The prefix `pvw_` and then every character a token is made of, however far that goes, rather
+ * than the exact 22.digits.22 shape. A truncated token is still part of the secret, a log line cut at 200
+ * characters is the ordinary way one gets truncated, and a looser match costs nothing: the marker replaces
+ * only what looked like a token. The token alphabet is base64url plus the two dots, or any of those characters
+ * written as a percent-escape, once or several times (`encodeURIComponent` of a URL that was itself in a query
+ * string). `packages/data-provider/specs/hostedPreview.spec.ts` pins this against the contract's token shape
+ * and `api/server/services/HostedPreview/__tests__/telemetryRedaction.spec.js` against tokens the issuer really
+ * mints, so a change to either side fails there rather than in a trace nobody reads.
+ *
+ * THE OTHER PATH CREDENTIALS (OSK-16174). Four more bearer credentials travel in a URL, and until they were added here
+ * only the request logger's path rewrite (`requestLogger.js`) knew them, so the same address that was scrubbed from a
+ * log line reached a trace and an error report in full: a workspace invitation (`wsi_`), a platform invitation
+ * (`pinv_`), a member's invitation link (`usr_`) and an operator's (`adm2_`, and the older `adm_`). Unlike `pvw_`, whose
+ * prefix is rare enough to match on its own, `usr_` and `adm_` begin ordinary words and identifiers (`/usr_local`,
+ * `adm_users`), so each of these is recognised by the SHAPE its minting code gives it, and a lookalike is left alone:
+ *
+ *   wsi_ / pinv_ / usr_ / adm2_   `<prefix>_<32 hex>.<version>.<key id>.<43 base64url>`, from `buildCredential` in
+ *                                 `api/models/WorkspaceInvite.js` and `PlatformInvitation.js` and `buildUserCode` and
+ *                                 `buildAdminCode` in `InvitationCode.js`. The older member link, `usr_<32 hex>.<24
+ *                                 base64url>`, has the same first two fields. Matched from the prefix to the end of the
+ *                                 token's characters once the 32-hex id and its dot are there, so a credential cut off
+ *                                 later in the log line is still redacted whole.
+ *   adm_                          `adm_<43 base64url>`, the older operator code (`generateCode`: 32 random bytes).
+ *                                 Exactly 43 characters, because that is the only length it has.
+ *
+ * `api/server/services/HostedPreview/__tests__/telemetryRedaction.spec.js` mints credentials with the real minting
+ * code and requires every one to come out as the marker, so a change to either side fails there.
+ *
+ * WHAT IT DOES NOT. A prefix that is itself percent-encoded (`%70vw_`) is not recognised: nothing in this
+ * repository writes one, and a scan that decoded every string first would cost every sink on every record. A
+ * credential cut off before its id is whole (`pinv_0123abc`) is an identifier, not yet the secret, and is left as it
+ * is. The request log's own `/[credential]` rewrite stays as it is, and is looser (any `wsi_`, `pinv_`, `usr_`,
+ * `adm_`, `adm2_` or `pvw_` path segment), because a path segment is a place a credential is known to be.
+ *
+ * COST. Linear in the input. The only nested repetition is `%(?:25)*XX`, and `XX` is never `25`, so the engine
+ * cannot read one run of `25`s two ways; the tests feed it a megabyte of the worst shapes.
+ *
+ * NEVER THROWS. These run inside telemetry hooks, and a hook that throws takes a request or an error report
+ * with it. Whatever cannot be scrubbed is reported to the caller as such (`false`), and the caller drops the
+ * record, because a record with the token still in it is exactly what this exists to prevent.
+ */
+/** What replaces a token. Fixed, so a trace can say a preview was involved but not which one. */
+export const PREVIEW_TOKEN_REDACTION = 'pvw_[redacted]';
+/**
+ * One character of a token: base64url, a dot, or any of those written as a percent-escape. An escape that was
+ * escaped again is `%25` followed by the original two digits (`_` is `%5F`, then `%255F`, then `%25255F`), which is
+ * what the `(?:25)*` is for. `25` is not itself a digit pair of a token character, so a run of them can only be
+ * read one way.
+ */
+const TOKEN_CHAR = '(?:[A-Za-z0-9._-]|%(?:25)*(?:2[DdEe]|3[0-9]|4[1-9A-Fa-f]|5[0-9AaFf]|6[1-9A-Fa-f]|7[0-9Aa]))';
+/**
+ * `pvw` then an underscore, which may be written `%5F` (or `%255F`, ...); then at least ONE token character. Not
+ * zero: the marker is `pvw_[`, and `[` is not a token character, so a string that was already redacted matches
+ * nothing and redacting twice changes nothing.
+ */
+const PREVIEW_TOKEN = new RegExp(`pvw(?:_|%(?:25)*5[Ff])${TOKEN_CHAR}+`, 'g');
+/**
+ * A prefix's underscore, which (like the token's own characters) may be written as a percent-escape, once or several
+ * times; and the dot that ends an id.
+ */
+const UNDERSCORE = '(?:_|%(?:25)*5[Ff])';
+const DOT = '(?:\\.|%(?:25)*2[Ee])';
+/**
+ * `wsi`, `pinv`, `usr` or `adm2`, an underscore, and then a lookahead that insists on what the minting code writes
+ * next: 32 lowercase hex characters and the dot that ends them. Everything the credential is made of follows, and the
+ * prefix is captured so the marker can say which kind it was. `usr_<40 hex>`, `usr_00000001` and `/usr_profile` have no
+ * dot after exactly 32 hex characters, so they match nothing.
+ */
+const SIGNED_CREDENTIAL = new RegExp(`(wsi|pinv|usr|adm2)${UNDERSCORE}(?=[0-9a-f]{32}${DOT})${TOKEN_CHAR}+`, 'g');
+/**
+ * The older operator code: `adm_` and exactly 43 base64url characters (32 random bytes), not followed by another one.
+ * A shorter or longer run of letters is a name, not this credential. The characters are the plain alphabet and not
+ * `TOKEN_CHAR`, whose dot would swallow the full stop at the end of a sentence.
+ */
+const LEGACY_OPERATOR_CODE = new RegExp(`(adm)${UNDERSCORE}(?=[A-Za-z0-9_-]{43}(?![A-Za-z0-9_-]))[A-Za-z0-9_-]{43}`, 'g');
+/** Cheap test before any of the expressions: a string with none of these cannot hold a credential. */
+const CREDENTIAL_HINTS = ['pvw', 'wsi', 'pinv', 'usr', 'adm'];
+const credentialMarker = (_match, prefix) => `${prefix}_[redacted]`;
+export function redactPreviewTokens(value) {
+    if (typeof value !== 'string' || !CREDENTIAL_HINTS.some((hint) => value.indexOf(hint) !== -1)) {
+        return value;
+    }
+    return value
+        .replace(PREVIEW_TOKEN, PREVIEW_TOKEN_REDACTION)
+        .replace(SIGNED_CREDENTIAL, credentialMarker)
+        .replace(LEGACY_OPERATOR_CODE, credentialMarker);
+}
+const DEFAULT_MAX_NODES = 50_000;
+const DEFAULT_MAX_DEPTH = 32;
+/**
+ * Redact every string inside an object graph, IN PLACE (an error report is rewritten far more cheaply than it is
+ * copied). Returns `true` when the whole graph was scrubbed and `false` when some of it could not be: past
+ * the bounds, frozen, behind a getter that throws. A `false` means the value may still hold a token and must not
+ * be sent. Binary payloads are skipped, not walked, and Error messages and stacks are scrubbed though they are
+ * not enumerable.
+ */
+export function redactPreviewTokensDeep(root, limits = {}) {
+    if (root === null || typeof root !== 'object') {
+        return true;
+    }
+    const maxNodes = limits.maxNodes ?? DEFAULT_MAX_NODES;
+    const maxDepth = limits.maxDepth ?? DEFAULT_MAX_DEPTH;
+    try {
+        const seen = new WeakSet();
+        const stack = [[root, 0]];
+        let visited = 0;
+        let complete = true;
+        const rewrite = (holder, key, current) => {
+            const next = redactPreviewTokens(current);
+            if (next === current) {
+                return;
+            }
+            try {
+                holder[key] = next;
+            }
+            catch {
+                // a frozen object throws in strict code
+            }
+            if (holder[key] !== next) {
+                // and silently keeps the old value in sloppy code
+                complete = false;
+            }
+        };
+        while (stack.length > 0) {
+            const [node, depth] = stack.pop();
+            if (seen.has(node) || ArrayBuffer.isView(node)) {
+                continue;
+            }
+            seen.add(node);
+            visited += 1;
+            if (visited > maxNodes) {
+                return false;
+            }
+            const holder = node;
+            if (node instanceof Error) {
+                for (const key of ['message', 'stack']) {
+                    if (typeof holder[key] === 'string') {
+                        rewrite(holder, key, holder[key]);
+                    }
+                }
+            }
+            for (const key of Object.keys(node)) {
+                let child;
+                try {
+                    child = holder[key];
+                }
+                catch {
+                    complete = false;
+                    continue;
+                }
+                if (typeof child === 'string') {
+                    rewrite(holder, key, child);
+                }
+                else if (child !== null && typeof child === 'object') {
+                    if (depth + 1 > maxDepth) {
+                        complete = false;
+                    }
+                    else {
+                        stack.push([child, depth + 1]);
+                    }
+                }
+            }
+        }
+        return complete;
+    }
+    catch {
+        return false;
+    }
+}
+function isThenable(value) {
+    return (value !== null &&
+        (typeof value === 'object' || typeof value === 'function') &&
+        typeof value.then === 'function');
+}
+/** The record with its tokens redacted, or `null` (dropped) when that could not be done. */
+function scrubRecord(record) {
+    if (record === null || record === undefined) {
+        return record;
+    }
+    if (typeof record !== 'object') {
+        return null;
+    }
+    return redactPreviewTokensDeep(record) ? record : null;
+}
+function guard(existing) {
+    return (value, hint) => {
+        const result = existing ? existing(value, hint) : value;
+        return isThenable(result) ? result.then(scrubRecord) : scrubRecord(result);
+    };
+}
+/**
+ * Sentry `init` options with a redacting `beforeSend`, `beforeSendTransaction` and `beforeBreadcrumb`. Those three
+ * are where an event's `request.url`, `request.query_string`, `transaction`, spans, breadcrumbs, `extra` and
+ * `contexts` pass last, so redacting the whole record there needs no list of fields and covers one added later.
+ * An existing hook runs FIRST and keeps its say (it can still drop the record, change it, or be asynchronous);
+ * what it returns is scrubbed after it, so nothing it adds escapes. The options object passed in is not modified.
+ * Every `Sentry.init` in this repository goes through this, and an architecture spec fails if one is added without.
+ */
+export function withPreviewTokenRedaction(options) {
+    const hooks = options;
+    return {
+        ...options,
+        beforeSend: guard(hooks.beforeSend),
+        beforeSendTransaction: guard(hooks.beforeSendTransaction),
+        beforeBreadcrumb: guard(hooks.beforeBreadcrumb),
+    };
+}

package/dist/node_modules/@skrr-ai/auth-core/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@skrr-ai/auth-core",
-  "version": "0.4.22",
+  "version": "0.4.25",
   "main": "dist/cjs/index.js",
   "types": "dist/esm/index.d.ts",
   "exports": {
@@ -183,6 +183,16 @@
         "types": "./dist/cjs/releasePolicy.d.ts",
         "default": "./dist/cjs/releasePolicy.js"
       }
+    },
+    "./telemetry-redaction": {
+      "import": {
+        "types": "./dist/esm/telemetryRedaction.d.ts",
+        "default": "./dist/esm/telemetryRedaction.js"
+      },
+      "require": {
+        "types": "./dist/cjs/telemetryRedaction.d.ts",
+        "default": "./dist/cjs/telemetryRedaction.js"
+      }
     }
   },
   "dependencies": {