[BUG] Hitting limits in 1 chat for multi-step healthcare compliance document workflows
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?
Max 20x subscriber. I use Claude for multi-step healthcare compliance document workflows — building Word docs, Excel risk registers, PowerPoint slides, and running file operations in a single session. These tasks routinely require 30–50 tool calls per turn. The current ~20 call cap turns a coherent autonomous workflow into a manual babysitting exercise that also increases my risk of hitting session limits faster due to compounding context. Specific request: plan-tiered limits with Max 20x at unlimited or configurable per-turn threshold.
What Should Happen?
Behavior shoudl mimic changes prior to March 23
Error Messages/Logs
Steps to Reproduce
Title: [BUG] Per-turn tool-use limit hit on claude.ai web — Max 20x plan — multi-file document generation workflowsEnvironment
Plan: Max 20x ($200/month)
Surface: claude.ai web interface (not Claude Code / terminal)
Model: Claude Sonnet 4.6
Platform: Web browser
Date first observed: March 29, 2026
Steps to Reproduce
Open a new conversation on claude.ai with the following tools enabled: Code Execution & File Creation, Web Search, Artifacts
Upload a ZIP file containing multiple document types (.docx, .xlsx, .pptx, .tf, .md files — 19 files total in this case)
Issue a single prompt requesting multi-file revision across all documents simultaneously — specifically: extract ZIP contents, read each file, apply branded text substitutions across all files, inject new content sections, repack files, validate output, and repackage into a new ZIP
Observe Claude begin chaining tool calls: file extraction → XML unpacking → document reading (pandoc) → patch scripting → repacking → validation → repeat across 5 Word documents + 2 Excel files + 3 Terraform README files
Tool-use limit is hit mid-task, typically after approximately 20–30 tool calls, before the workflow completes
The "Continue" button resumes but breaks autonomous execution — Claude occasionally loses track of which files have been processed, requiring re-orientation
Expected BehaviorOn a Max 20x plan, a well-specified multi-file document workflow should complete autonomously without per-turn interruption. The session had ample usage budget remaining. The tool-use limit is a separate, lower-level cap that does not scale with plan tier and fires regardless of remaining session quota.Actual Behavior"Claude reached its tool-use limit for this turn" banner appears mid-workflow. Manual Continue click required. On complex tasks, resumption sometimes causes Claude to re-read already-processed files, consuming additional tool calls and session quota unnecessarily.
Claude Model
None
Is this a regression?
Yes, this worked in a previous version
Last Working Version
_No response_
Claude Code Version
Sonnet 4.6 extended
Platform
Anthropic API
Operating System
macOS
Terminal/Shell
Terminal.app (macOS)
Additional Information
I was trying to rework our ISO 27001 documents to cater to upcoming proposed AI regulations and NIST guidelines. I uploaded less than 30- docs and hit the limit.
This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗