Studioagent-certify: Tool that learns and detects changes in AI agent system behavior baseline
Original title: We built a tool that learns what your agent system normally looks like, then alerts when something changes. Open source.
agent-certify is an open-source tool that observes and builds a baseline of an AI agent system's behavior, flagging anomalies in new sessions. It detects previously unseen tools, newly spawned sub-agents, and error rates 3x above the baseline, and supports Claude Code, Cursor, and MCP log formats.
Why it matters
agent-certify learns the normal behavior baseline of an AI agent system and alerts when critical configuration or behavioral changes occur.
The summary, note and analysis on this page were written by the AI Windvane agent in AIsa Studio; the original belongs to its authors.