Issue Station
Sign In
Tools hereToolbox

claude-code - 💡(How to fix) Fix System prompt size tripled after update to v2.1.101, causing 3x faster quota consumption [1 comments, 2 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
anthropics/claude-code#48050•Fetched 2026-04-15 06:34:43
View on GitHub
Comments
1
Participants
2
Timeline
7
Reactions
0
Author
rbayleyrbayley
Participants
raye-dengraye-dengrbayleyrbayley
Timeline (top)
labeled ×4commented ×1cross-referenced ×1subscribed ×1

After updating Claude Code from v2.1.71 to v2.1.101, the base system prompt size tripled, causing usage quota to be consumed ~3x faster than before.

Root Cause

This is consistent across multiple projects, confirming it's not caused by project-specific CLAUDE.md files or integrations.

RAW_BUFFERClick to expand / collapse

Summary

After updating Claude Code from v2.1.71 to v2.1.101, the base system prompt size tripled, causing usage quota to be consumed ~3x faster than before.

Evidence

Comparing cache_read_input_tokens from conversation JSONL files before and after the update:

VersionDatecache_read_input_tokens
2.1.71April 6, 2026~54,000
2.1.101April 14, 2026~163,000

This is consistent across multiple projects, confirming it's not caused by project-specific CLAUDE.md files or integrations.

Impact

  • Each conversation turn now consumes ~2% of the hourly quota just from the base system prompt overhead
  • Previously the same usage patterns would consume roughly 1/3 of that
  • Affects all projects regardless of configuration

What was ruled out

  • No active MCP integrations (Google Calendar, Gmail, etc. — all disconnected)
  • No large CLAUDE.md files in global ~/.claude/ directory
  • No enabled plugins injecting content
  • Consistent across different projects

Environment

  • OS: Windows 11
  • Previous version: 2.1.71
  • Current version: 2.1.101
  • Model: claude-sonnet-4-6

extent analysis

TL;DR

Downgrade to version 2.1.71 or investigate the cause of the increased cache_read_input_tokens in version 2.1.101 to mitigate the quota consumption issue.

Guidance

  • Review the changelog for version 2.1.101 to identify potential changes that could be causing the increase in cache_read_input_tokens.
  • Compare the configuration and settings between versions 2.1.71 and 2.1.101 to ensure no unintended changes were made.
  • Investigate the cache_read_input_tokens metric to understand what contributes to its value and how it can be optimized.
  • Consider reaching out to the Claude Code support team for assistance in identifying the root cause of the issue.

Example

No code snippet is provided as the issue does not involve code changes.

Notes

The increased cache_read_input_tokens value is consistent across multiple projects, suggesting a systemic issue with version 2.1.101. However, without further information about the changes made in version 2.1.101, it is difficult to provide a more specific solution.

Recommendation

Apply workaround: Downgrade to version 2.1.71 until the root cause of the issue is identified and resolved, as this version did not exhibit the increased quota consumption issue.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#prompt issue#agent setup#task chaining#parallel task#integration issue

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING