Check Point Research found that a cloned repository's Claude Code configuration could run hooks, start MCP servers before the user approved them, and redirect API traffic so the user's Anthropic API key was sent to an attacker (CVE-2025-59536, CVE-2026-21852). Anthropic fixed the issues between August and December 2025 by deferring execution and API calls until after the trust dialog.
Why it matters
Repository-level agent configuration is executable attack surface that triggers when a developer simply opens a project.
Key facts
As stated in the sources, with where to find them.
- Hooks issue reported 2025-07-21 and fixed 2025-08-26 (advisory GHSA-ph6w-f82w-28w6); MCP consent bypass reported 2025-09-03, fixed 2025-09-22, CVE-2025-59536 published 2025-10-03; API-key exfiltration reported 2025-10-28, fixed 2025-12-28, CVE-2026-21852 published 2026-01-21.Timeline and Disclosure
Findings that cite this record
Key questions this bears on
- Where are deployed AI agents actually being exploited?Exploitation often reaches deployed agents through connectors, credentials, tools and packages.
- Do cyber evaluations of AI agents stay contained?Not reliably. Agents have reached real systems from evaluation, training and research; red-teaming found containment gaps OpenAI says it largely fixed.
- Can AI agents defend and oversee systems on their own?Not reliably. Some designs improve benchmark triage, but autonomous defense lacks operational validation and production oversight of agents has documented gaps.
Sources
Related records
Apr 15, 2026
Aug 5, 2025
Oct 26, 2026
Sep 30, 2026
Apr 1, 2025
Mar 30, 2025