[ctx]ctxseccontext security: attacks on the harness and context layer of AI agents, plus exploited-now news, hunting and playbooks.

← Cybersecurity News · markdown

official (VulnCheck CNA advisory, Nov 13, 2026; GitHub advisory GHSA-pw3r-7vg3-gqch); CVSS 4.0 9.4 critical; no fixed release listed; not reported exploited

Tactical RMM CVE-2026-129165: the report preview endpoint reads any server file, exposing the Django SECRET_KEY that doubles as the NATS password and lets a report manager run commands on every managed agent

  • Affected: Tactical RMM 0.17.0 through 1.6.1, endpoint /reporting/templates/preview/ (reporting is in the ee code path). Credit: Wu Yulun.
  • Precondition: an account with report management permission (CVSS PR:H), but the result is command execution on every endpoint the RMM manages.
  • Status: no fixed release named in the advisory as of Nov 14; watch the amidaware/tacticalrmm releases page. Not reported exploited.
  • Until a fix (our suggestion, not from the advisory): limit report management permission to as few admins as possible, review who holds it, and if you suspect misuse treat the SECRET_KEY/NATS credential as exposed and rotate it.

Sources: VulnCheck advisory: Tactical RMM 0.17.0 through 1.6.1 Arbitrary File Read via Report Preview (CVE-2026-129165) · GitHub Advisory Database: GHSA-pw3r-7vg3-gqch · Tactical RMM releases