[BUG] Session reaches 100% usage with no active Claude processes, subagents, or background commands
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code
What's Wrong?
Claude Code reports the current session as 100% used despite no active Claude-related activity on the machine.
I initially suspected that usage was caused by large-context sessions, MCP integrations, connectors, subagents, or background commands. After investigating, I found evidence that these features had been used historically.
However, after cleaning the environment and performing a forensic audit, I found:
- No active Claude processes
- No active Anthropic processes
- No active browser bridge
- No Claude-related Windows services
- No Claude-related scheduled tasks
- No active subagents in the last 12 hours
- No active background commands in the last 12 hours
Despite this, Claude Code still reports:
- Current session: 100% used
- Local usage: 0 input, 0 output, 0 cache read, 0 cache write
- API duration: 0s
The usage report attributes consumption to historical subagent-heavy sessions, but the last identified subagent execution completed successfully on 2026-06-05 and no activity was found afterward.
It is unclear whether this behavior is expected, delayed accounting, or a bug in session usage reporting.
What Should Happen?
Claude Code should accurately reflect current session usage.
If historical subagent activity is responsible for the session being exhausted, the UI should clearly indicate that the usage comes from previously completed work.
If there is no active work occurring, users should be able to determine why a session remains at 100% usage.
Error Messages/Logs
Session
Total cost: $0.0000
Total duration (API): 0s
Total duration (wall): 26s
Total code changes: 0 lines added, 0 lines removed
Usage: 0 input, 0 output, 0 cache read, 0 cache write
Current session
100% used
100% of your usage came from subagent-heavy sessions
27% of your usage came from subagents under "Explore"
Steps to Reproduce
- Use Claude Code with subagents (Explore) and large-context sessions.
- Allow all subagents and background commands to complete.
- Close Claude Code and stop using it for an extended period.
- Verify that no Claude-related processes, services, scheduled tasks, or browser bridges remain active.
- Reopen Claude Code and run
/usage. - Observe that the current session may still report 100% usage despite showing:
- 0 input tokens
- 0 output tokens
- 0 cache reads/writes
- 0 API duration
Claude Model
Sonnet (default)
Is this a regression?
I don't know
Last Working Version
_No response_
Claude Code Version
2.1.167 (Claude Code)
Platform
Other
Operating System
Windows
Terminal/Shell
Terminal.app (macOS)
Additional Information
Forensic audit performed:
- No active Claude processes
- No active Anthropic processes
- No active browser bridge
- No Windows services related to Claude
- No scheduled tasks related to Claude
- Last subagent execution:
2026-06-05 13:21:53 -> 13:23:47
- No subagent activity in the last 12 hours
- No background command activity in the last 12 hours
Historical activity was found, but no evidence of ongoing local activity was found during the period where session usage remained exhausted.
This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗