1 comments

  • vchain-dz 6 hours ago ago

    I built AgentMon because I use Claude Code and Codex in different terminals and needed a way to monitor all my sessions, switching models and the done gate. No data leaves your system, all run on localhost and is restricted by default.

    It started as a Claude Code mod with a pane beside the chat. Now it has a local browser view, that supports also Codex and Kiro Crew sessions. AgentMon Live also provides insights to optimize the agent sessions.

    There is a Claude-specific completion check, that before Claude finishes, checks for things like failed checks, missed tests, no-test runs and unfinished tasks. When detected it can send unfinished work back to Claude once.

    Many new features are based on Reddit community feedback, and there is a lot of day to day experience built-in.

    The repo has installation guidelines, screenshots and a demo. Project is Apache-2.0 licensed.

    I'm super interested in feedback to make it the most useful for monitoring agentic coding. If you like it you can also star the repo as a sign.