Each feed is checked for new posts. Every post is read, and those that belong here join the catalog. What Trove read and did not keep stays on file, not shown.
Tenet Threat Labs red-teamed a GPT-4o agent with all five A2AS BASIC pillars enabled and an MCP server. A tool response disguised as HTTP 206 pagination and OAuth steps ("Tool Parameter Injection via Semantic Orchestration Mimicry") led the agent to run a shell script, read credentials.txt and exfiltrate the token. It succeeded in over 80% of runs in November 2025. The authors argue for runtime enforcement based on content trust levels.
Model Context ProtocolGPT-4oA2AS
Tool Parameter Injection via Semantic Orchestration Mimicryprompt injectionindirect prompt injection+1
Agentjacking is an attack by Tenet Security researchers in which a fake bug report or poisoned log makes an AI coding agent run attacker code during a routine triage request. They report an 85% success rate against Cursor, Claude Code and Codex, including network-disabled sandboxes, and confirmation across 2,388 organizations. The recording covers sandbox and container bypasses and defenses, including the open-source JackStop hardening tool.
tenetsecurity.aiwriteup