You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The footer progress bar for dynamic workflows (Workflow tool / /workflows) is misleading, and its layout keeps changing between releases. As a user I only need three stable things: what is running now, a reliable completion indicator, and an estimated finish time.
Environment
Claude Code 2.1.284, macOS 15.7.7, terminal CLI, auto mode on
Two workflows running at the same time in one session
What happens
The bar counts finished agents only. A workflow with 2 sequential agents (fix, then review) shows 0 % for ~50 minutes while its first agent is actively working (233k tokens so far), then jumps.
A workflow whose agent count depends on results (an optional retry round) has no stable denominator, so the bar can also move backwards or stall.
The only per-workflow figure is a token counter (e.g. ↓ 233.4k, ↓ 2.3m), which does not tell whether the run is progressing or stuck.
A simple, stable one-line-per-workflow view (the current compact style is fine), with:
Current step: active phase and agent label (e.g. Revue · agent 2/2).
Reliable progress: count running agents as partially done, or show done / running / queued explicitly, instead of a bar stuck at 0 %.
Elapsed time per workflow and for the active agent, plus time since its last activity (to spot a hang).
ETA when estimable (e.g. from completed agents' durations in the same run or from previous runs of the same script), shown as a range, or omitted, never a fake precise time.
A stable layout across releases, or a documented setting to keep the compact view.
Steps to reproduce
Launch a workflow with two sequential agent() calls, where the first one runs for 30+ minutes.
Watch the footer: the bar stays empty until the first agent returns.
Summary
The footer progress bar for dynamic workflows (Workflow tool /
/workflows) is misleading, and its layout keeps changing between releases. As a user I only need three stable things: what is running now, a reliable completion indicator, and an estimated finish time.Environment
What happens
↓ 233.4k,↓ 2.3m), which does not tell whether the run is progressing or stuck.Expected
A simple, stable one-line-per-workflow view (the current compact style is fine), with:
Revue · agent 2/2).done / running / queuedexplicitly, instead of a bar stuck at 0 %.Steps to reproduce
agent()calls, where the first one runs for 30+ minutes.