29-Year-Old Squid Proxy Bug ‘Squidbleed’ Leaks Cleartext HTTP Requests

A critical heap over-read vulnerability in the Squid web proxy, dubbed ‘Squidbleed’ (CVE-2026-47729), can leak cleartext HTTP requests—including credentials and session tokens—to other users of the same proxy. The bug originates from a 1997 FTP-parsing change and remains active in Squid’s default configuration.

Disclosed by researchers at Calif.io in June 2026, Squidbleed affects shared network environments such as schools, offices, and public Wi-Fi, where an attacker is already a trusted client. The vulnerability only exposes traffic that Squid can read; normal HTTPS traffic via CONNECT tunnels remains opaque unless Squid performs TLS termination and inspection.

The flaw resides in Squid’s FTP directory-listing parser. When an attacker-controlled FTP server sends a listing line ending immediately after the timestamp with no filename, the parser’s whitespace-skipping loop fails to stop at the null terminator, causing an out-of-bounds read. Squid’s reuse of unzeroed memory buffers then returns remnants of previous HTTP requests, including sensitive data like Authorization headers.

Proof-of-concept code is publicly available, though no in-the-wild exploitation has been reported. The fix involves adding a null-terminator check before the vulnerable strchr calls, merged into Squid’s development branch in April and v7 in May. However, version confusion exists: maintainer Amos Jeffries initially stated Squid 7.6 carried the fix, then corrected to 7.7, while Debian’s Salvatore Bonaccorso noted the referenced commit appears in 7.6. SUSE rates the vulnerability as moderate (CVSS 6.5), with only confidentiality impact.

Researchers recommend disabling FTP support in Squid entirely, as most networks carry little FTP traffic. Calif.io credits Anthropic’s Claude Mythos Preview (part of Project Glasswing) for rapidly identifying the strchr quirk, highlighting AI’s growing role in uncovering buried parser bugs.