CVE-2026-78379
Improper neutralization of input used for LLM prompting in the python_repl tool in Amazon Strands Agents Tools before 0.8.5 might allow remote actors to execute arbitrary Python code on the agent's host by bypassing the human consent gate, via a crafted prompt that forwards non_interactive_mode as a keyword argument through the batch tool. To remediate this issue, users should upgrade to version 0.8.5 or later.
CSIRTS triage
- What
- A consent bypass vulnerability in the python_repl tool allows remote attackers to execute arbitrary Python code by bypassing the human approval gate through crafted LLM prompts.
- Who is affected
- AI agents using Strands Agents Tools python_repl before version 0.8.5 that process untrusted LLM-generated input.
- Urgency
- High; allows arbitrary code execution on agent hosts via prompt injection without user consent.
- Action
- Upgrade strands-agents-tools to version 0.8.5 or later.
AI-assisted analysis generated from the source advisory — verify against the original.
⚡ Watch CVE-2026-78379
Get an email if CVE-2026-78379 is added to CISA KEV, gains public exploit code, or a new advisory cites it — max one per day, one-click unsubscribe.
Advisory coverage (2)
External references
Embed the live status
— this badge updates automatically when the KEV or exploit status changes. How to embed it →
[](https://www.csirts.com/cve/CVE-2026-78379)