claude-code - 💡(How to fix) Fix [Bug] Excessive token consumption during model initialization or analysis [1 comments, 2 participants]

[{"error":"Error: ENOENT: no such file or directory, unlink 'C:\Users\ben\AppData\Local\Temp\claude\C--Users-ben\54b34086-2767-4ace-9abb-28133e5dbfda\tasks\a1ca351493370355f.output'\n at async <anonymous> (B:/~BUN/root/src/entrypoints/cli.js:7979:299)","timestamp":"2026-04-21T02:18:45.259Z"},{"error":"Error: ENOENT: no such file or directory, unlink 'C:\Users\ben\AppData\Local\Temp\claude\C--Users-ben\54b34086-2767-4ace-9abb-28133e5dbfda\tasks\a53a454aaeaddfcd3.output'\n at async <anonymous> (B:/~BUN/root/src/entrypoints/cli.js:7979:299)","timestamp":"2026-04-21T02:20:02.239Z"},{"error":"Error: ensureToolResultPairing: repaired missing tool_result blocks (7 -> 7 messages). Message structure: [0] user; [1] assistant(id=msg_01SstcUXaU2Rn57qRUvzLfPx, tool_uses=[toolu_01AtNtN9sSi46nsxjUGWxf7g]); [2] user(tool_results=[toolu_01AtNtN9sSi46nsxjUGWxf7g]); [3] assistant(id=msg_01979teX2TCLpoWtPt2Jvq12, tool_uses=[toolu_015EeBK5CWKN4ZbzioWmErEe]); [4] user(tool_results=[toolu_015EeBK5CWKN4ZbzioWmErEe]); [5] assistant(id=msg_012Sc8QvVz7r82UVfQGvm7uw, tool_uses=[toolu_01CPXWxD9n9sp6BogwCpX4WT]); [6] user\n at C0_ (B:/~BUN/root/src/entrypoints/cli.js:5285:10250)\n at Qi_ (B:/~BUN/root/src/entrypoints/cli.js:8129:16029)\n at processTicksAndRejections (native:7:39)","timestamp":"2026-04-21T02:20:44.969Z"}]

TL;DR

The issue is likely due to excessive token consumption by Claude's exploratory agents, and a potential workaround is to limit the scope of the conversation or investigate the errors causing the agents to fail.

Guidance

• Investigate the ENOENT errors in the error log, which may indicate issues with file access or temporary directory management, potentially contributing to the excessive token consumption.
• Review the conversation history and background information to determine if there are any specific topics or keywords that may be triggering the exploratory agents, and consider limiting the scope of the conversation to reduce token usage.
• Check the tool usage and results in the conversation to ensure that they are necessary and not causing unnecessary token consumption.
• Consider reporting the issue with the provided Feedback ID to the development team for further investigation and potential optimization of the token usage.

Example

No specific code snippet can be provided without further information on the conversation or the Claude model being used.

Notes

The issue may be specific to the Windows platform or the version of the software being used, and further investigation is needed to determine the root cause of the excessive token consumption.

Recommendation

Apply a workaround by limiting the scope of the conversation and investigating the errors causing the exploratory agents to fail, as the root cause of the issue is unclear and may require further investigation by the development team.