Vulnerability research write-ups. Every entry here corresponds to a fixed, publicly tracked issue — no theoretical findings.
Adel Bouachraoui — hydrasecurity.fr/v10/ · LinkedIn · YesWeHack
| Write-up | Target | Class | ID |
|---|---|---|---|
| When a security patch guards the wrong branch | GNOME libsoup | Out-of-bounds read (CWE-125) | CVE-2026-12478 |
Further write-ups are held until the corresponding issue is fixed and disclosure is cleared through the relevant program.
- CVE-2026-28962 — Apple WebKit. Malicious web content leading to sensitive-information disclosure. Fixed in Safari 26.5, macOS, iOS/iPadOS and visionOS. (Apple advisory)
Source-driven auditing over blind fuzzing. Two things I look for in particular:
- Incomplete fixes. A patch is written under time pressure by someone who has already decided what the bug is. That framing is where the second bug lives.
- Guards placed inside conditionals. A bounds check is only as good as the set of paths that reach it. Protocol parsers with distinct handling per direction, per opcode or per frame type are structurally prone to this.
Nothing is a finding until an exploit oracle fires. Severity is reported honestly, including when preconditions make a bug hard to reach.