Is my AI agent stuck retrying?

ACTVT watches the tool calls in each live agent session. If one tool fails at least 5 times within 10 minutes, and most of that tool’s calls are failing, it sends a notification like “A tool keeps failing in <session>” or “Bash keeps failing”. You get it once per tool per run. Open the session to see each failed call, then decide whether to stop it, fix whatever it depends on, or change what you asked it to do. ACTVT only counts failures. It can’t tell whether the agent is getting anywhere. Repeated tool failures is on by default (Settings → Notifications).

0:384 chapters

Step by step

  1. 0:00You ask

    I ask my agent to upload a file. The server can be flaky, so I tell it to “try again until it goes through”.

  2. 0:06It keeps trying

    But the server is actually down, so every upload fails and it just keeps trying.

  3. 0:14ACTVT tells you

    ACTVT lets me know that the same tool has failed 5 times within 10 minutes.

  4. 0:22See what happened

    When I open the session, there’s my instruction and then one attempt after another. So far that’s 11 failed calls and $0.06 at API rates. It isn’t going to stop on its own, so I can stop it, fix the server, or change the instruction. This notification is on by default.

Try it on your own agents

ACTVT is free to start, and reads the session files Claude Code, Codex CLI and Pi already write. Nothing to configure.