
The runtime engine
Sentinel for your AI agents
The engine that powers Agentbeam. It records shell commands, file edits, tool calls, and network activity from your coding agents into a local timeline, flags risky actions with the evidence behind them, and scans MCP configs and SKILL.md files before you trust them.
- Runtime activity recorded as local NDJSON files — no cloud service, no telemetry
- 11 heuristic patterns plus a version-pin check for tool poisoning, rug pulls, and credential exfiltration
- Works across Claude Code and other terminal-based or IDE agents
