Runtipi is a personal homeserver orchestrator. In 4.10.0 and earlier, Runtipi accepts symbolic links from an attacker-controlled backup archive and copies them into live application paths during the backup restore flow. An authenticated attacker can plant user-config/app.env as a symlink to an arbitrary reachable path and then send PUT /api/user-config/demoapp3:_user with attacker-controlled appEnv content. FilesystemService.writeTextFile() follows the planted link, allowing content to be written outside the intended restore and user-config directory boundary with Runtipi process permissions. This issue is fixed in version 4.10.1.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
1 valid exploit after Mallory filtered fakes, detection scripts, and README-only repos.
Repository is a small standalone Python proof-of-concept for CVE-2026-55168 affecting Runtipi. It contains two files: a README describing the vulnerability and usage, and a single executable script, poc.py, which is the exploit entry point. The exploit is authenticated and web-based: it logs into a target Runtipi instance using supplied credentials, builds an in-memory tar archive compressed with gzip, and inserts a symlink entry user-config/app.env pointing to an attacker-chosen filesystem path (default /data/state/proof.txt). It then uploads this archive to the backup endpoint for a chosen app URN, triggers the restore flow, waits 8 seconds, and finally sends a normal user-config update request with attacker-controlled appEnv content. Because the restored symlink is followed during the later write, the script achieves arbitrary file write outside the intended application directory. The code uses Python standard libraries only (argparse, tarfile, gzip, urllib, cookiejar, json, re, time), maintains session cookies after login, and sets Origin/Referer/X-Forwarded-* headers to resemble expected application traffic. This is a real exploit PoC rather than a detector: it performs the full exploitation chain and demonstrates successful compromise by writing controlled content to a target file.
No public activity tracked yet. Mallory keeps watching.
No public activity observed for this vulnerability.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.