* [Agenta v0.113.0](https://github.com/Agenta-AI/agenta) – LLMOps platform offering integrated prompt playground, prompt management, evaluation, and observability for building reliable LLM applications.

* [TraceRoot v0.4.0](https://github.com/traceroot-ai/traceroot) – Observability platform that captures LLM calls, agent actions, and tool usage and provides AI-driven root-cause debugging tied to source and Git history.

* [managed-agents v0.3.0](https://github.com/sandbaseai/sandbase-harness) – Local-first runtime for AI agents with persistent sessions, sandboxed tools, SQLite-backed storage, audit/replay, and a built-in console UI.

* [ax v0.43.0](https://github.com/Necmttn/ax) – Local agent-experience loop that learns from AI coding agent sessions, records retro experiments in a typed graph, and proposes repo-specific fixes for review.

* [agent-inspect @agent-inspect/studi...](https://github.com/rajudandigam/agent-inspect) – Local evidence debugger for TypeScript AI agents that turns runs into execution trees for debugging, regression checks, and share-checked offline artifacts.