[FEATURE] Show timestamps next to messages/tool calls in terminal output
Preflight Checklist
- [x] I have searched existing requests and this feature hasn't been requested yet
- [x] This is a single feature request (not multiple features)
Problem Statement
When running long tasks (multi-step workflows, TDD loops, subagents), there's no way to tell when individual steps started or how long they took.
showTurnDuration only reports total turn time, which doesn't help identify which specific step was slow. Scrolling back through a 20-minute agent run, everything looks the same — no temporal context.
Proposed Solution
Add a setting like showTimestamps (off by default) that prepends a local timestamp [HH:MM:SS] to each message/tool invocation line in the terminal output. This gives per-step timing without changing the default experience.
Alternative Solutions
- Show elapsed time since session start instead of wall clock (e.g., [+5:32])
- Show duration per step after completion (e.g., "⎿ Done (2m 47s)") rather than start time
- Only show timestamps in verbose view mode
Priority
Medium - Would be very helpful
Feature Category
Configuration and settings
Use Case Example
Current output:
⏺ Running the full test suite.
⏺ Bash(vendor/bin/phpunit)
⎿ OK (142 tests, 298 assertions)
⏺ All tests pass. Running static analysis.
⏺ Bash(vendor/bin/phpstan analyse src/)
⎿ [OK] No errors
⏺ Starting code review.
⏺ Skill(check)
⎿ Done
With timestamps:
⏺ [14:32:05] Running the full test suite.
⏺ [14:32:06] Bash(vendor/bin/phpunit)
⎿ OK (142 tests, 298 assertions)
⏺ [14:37:42] All tests pass. Running static analysis.
⏺ [14:37:43] Bash(vendor/bin/phpstan analyse src/)
⎿ [OK] No errors
⏺ [14:38:15] Starting code review.
⏺ [14:38:16] Skill(check)
⎿ Done
Now it's obvious the test suite took ~5.5 minutes while static analysis took ~30 seconds.
Additional Context
Particularly useful when reviewing background agent output after the fact, or when running multiple agents in parallel and comparing their efficiency. The existing showTurnDuration setting shows this is already a recognized need — this extends it to per-step granularity.
This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗