Live agent network for Claude - know what your agents are acutally doing
What it does
Claude Code can spin up subagents, run dozens of tool calls and finish with a one-line summary. trovu shows what happened in between. A small hook sends each tool call and subagent event to your trovu account. You see: Agent network: the main session and every subagent with its task, call count and result, live. Active agents pulse green. Timeline and lanes: every call in order, per agent, with filters, search and an errors-only switch. Activity: all reported events across sessions, with durations and failures. Delete single sessions or events whenever you want to clean up. 200+ agents in one place: ready-made agent templates Claude can read over MCP while you are connected to trovu. Your own agents are tracked next to them. How it works: add the hook to Claude Code, connect Claude over MCP, and your sessions show up. trovu shows what the hook reports. It cannot verify it and it cannot stop Claude Code, so treat it as a debugging aid, not proof. Secrets that look like tokens or passwords are removed before anything is stored, and tool outputs and file contents are never stored.
Building something too? List it free — or give it a Sponsored spot.
1 comment
Sign in to comment.
I run my work with a team of Claude Code agents and kept hitting the same wall: subagents make dozens of tool calls, and you get a one-line summary. What happened in between is gone. trovu shows it: one hook sends every tool call and subagent event to a live agent network, timeline and activity feed. Useful when a run goes wrong, or before you trust a long session. It also includes my 200+ agents, which Claude can read over MCP. Let me know what you think!
0