{"version":"2.1.304","anchor":"workflow-agent-stall-retries-now-emit-a-telemetry-event","canonical_anchor":"workflow-agent-stall-retries-now-emit-a-telemetry-event","heading":"Retries of stalled workflow agents are now measured","tier":"internal","area":"Workflows","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.304\/e\/workflow-agent-stall-retries-now-emit-a-telemetry-event","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.304","markdown":"### Retries of stalled workflow agents are now measured\n\nWhen a workflow agent stalls and is retried, Claude Code now reports the restarts, the outcome and how much work was lost\n\n**What**\n\nA workflow runs several agents, which are helper copies of Claude with set jobs. When one of them stalls, loses its reply, or is retried by you, Claude Code now sends a usage report once the retrying is over. It records:\n\n- how many times the agent was restarted\n\n- how many attempts were made for each kind of problem\n\n- how it ended\n\n- how much context, tool calls and time were lost\n\nContext here means the conversation the agent had built up, and a tool call is one action Claude takes, such as running a command. The way Claude Code keeps track of these retries was also reworked.\n\n**Why**\n\nRetries of stalled workflow agents are now measured, including how much work each retry throws away.\n\n- Area: Workflows\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 2\/5\n- Scope: individual\n- Heads-up: no"}