I built Trackline, an open-source alignment and reliability layer for AI agents. It monitors whether agents are still acting within the task, rules and evidence they were given, locally with Claude Code, Codex and Cursor, and in production through OpenTelemetry traces.
I built Trackline, an open-source alignment and reliability layer for AI agents. It monitors whether agents are still acting within the task, rules and evide...