agentsight

Warn

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: MEDIUMPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PRIVILEGE_ESCALATION]: The skill requires root access to function (eBPF monitoring) and provides instructions for using sudo systemctl to manage the underlying service.
  • [COMMAND_EXECUTION]: The skill performs actions by executing the /usr/local/bin/agentsight binary with various flags to query system and token data.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests audit data from local SQLite databases (agentsight.db, genai_events.db). Since this data includes LLM call histories (inputs and outputs), it presents a surface for indirect prompt injection if the agent processes the logged content without sanitization.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 14, 2026, 03:58 AM