Vulnerabilities

OpenClaw Dashboard v3.0.0 contains a stored cross-site scripting vulnerability (CWE-79) that lets an unauthenticated remote attacker inject arbitrary HTML and script through the username field of a failed login POST request. The injected value is stored verbatim in the audit log and later rendered unescaped via innerHTML in the administrator's notification panel, where a permissive Content-Security-Policy permits inline event handlers, so the payload executes when an administrator opens the panel. Because the script runs in the administrator's session, the attacker can invoke authenticated endpoints — including editing agent instruction files and changing configuration — and, per the public proof-of-concept, achieve administrator account takeover. Anyone running OpenClaw Dashboard v3.0.0 whose login and notification panel are reachable to unauthenticated users is affected. The flaw is not yet in CISA's KEV and EPSS assigns roughly a 0.5% probability of exploitation within 30 days (42nd percentile), but two public proof-of-concept references are available, so opportunistic exploitation is plausible.

Do: Upgrade from v3.0.0 to a patched release as soon as the vendor publishes one (no fixed version is named in the disclosure); in the meantime, restrict the dashboard login page and notification panel to trusted networks or IP allowlists. Review stored audit-log entries for injected HTML/script content, and check whether agent instruction files or configuration were modified through any admin sessions. Defensively, ensure log fields are HTML-escaped before rendering and tighten the CSP to disallow inline event handlers.

LuaTeX before 1.17.0 allows execution of arbitrary shell commands when compiling a TeX file obtained from an untrusted source. This occurs because luatex-core.lua lets the original io.popen be accessed. This also affects TeX Live before 2023 r66984 and MiKTeX before 23.5.

LuaTeX before 1.17.0 allows a document (compiled with the default settings) to make arbitrary network requests. This occurs because full access to the socket library is permitted by default, as stated in the documentation. This also affects TeX Live before 2023 r66984 and MiKTeX before 23.5.

An issue was discovered in t1_check_unusual_charstring functions in writet1.c files in TeX Live before 2018-09-21. A buffer overflow in the handling of Type 1 fonts allows arbitrary code execution when a malicious font is loaded by one of the vulnerable tools: pdflatex, pdftex, dvips, or luatex.

TeX Live through 20170524 does not validate strings before launching the program specified by the BROWSER environment variable, which might allow remote attackers to conduct argument-injection attacks via a crafted URL, related to linked_scripts/context/stubs/unix/mtxrun, texmf-dist/scripts/context/stubs/mswin/mtxrun.lua, and texmf-dist/tex/luatex/lualibs/lualibs-os.lua.

TeX Live allows remote attackers to execute arbitrary commands by leveraging inclusion of mpost in shell_escape_commands in the texmf.cnf config file.