Glean CVE-2026-54339

Requires an authenticated Glean account able to reach POST /api/feeds/discover (PR:L) on a version prior to 0.2.6; no user interaction and no special non-default configuration are needed because feed discovery and background full-text extraction are core features. Exploitation succeeds against any deployment where the Glean backend/worker has network reachability to internal, loopback, link-local (169.254.0.0/16), or cloud-metadata (169.254.169.254) endpoints. The limiting factors are: valid authentication is mandatory (not anonymous), and impact is bounded to what the server itself can reach; disclosure is retrieved out-of-band via GET /api/entries/{id}, so the attacker needs read access to their own stored entries (which they have by default).

This is a genuine, well-substantiated SSRF (CWE-918) with above-average real-world impact because it is NON-BLIND: the fetched internal response is stored in Entry.content and readable via GET /api/entries/{id}, giving full response disclosure rather than blind timing/error signals. The CVSS 3.1 vector (AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:N/A:N = 7.7) is internally consistent with the description: network-reachable, low complexity, requires a valid authenticated account (PR:L) to POST /api/feeds/discover, scope-changed (S:C) because the server pivots to distinct internal systems, and high confidentiality with no integrity/availability impact. No EPSS score, KEV listing, POC, or SSVC data was provided in the input, so exploitation-probability signals are absent - per the precision rule this is 'no public exploit identified at time of analysis.' The patch commit (cdc05dd) is confirmed and introduces a self-contained SSRF-safe httpx client (safe_async_client) that blocks RFC1918/loopback/link-local/cloud-metadata ranges and resolves DNS-rebinding, corroborating both the root cause and the remediation. Net: a real, patch-now priority for anyone running Glean with the feed-discovery API exposed to untrusted authenticated users, but not evidence of mass exploitation - treat as targeted-risk given the product's small footprint.

An authenticated Glean user submits POST /api/feeds/discover with feed_url pointing to a malicious server that returns an RSS feed whose item links reference internal targets (e.g. http://169.254.169.254/latest/meta-data/ or http://127.0.0.1:PORT). The background worker fetches each link via fetch_and_extract_fulltext() without network validation, stores the response in Entry.content, and the attacker reads the full internal response through GET /api/entries/{id}. No public exploit code was identified in the provided data, but the attack is low-complexity and non-blind.