What this error means
MCP connector model intermittently emits tool_use instead of mcp_tool_use for tools declared in mcp_toolset — conversation stalls on client side is a Anthropic API failure pattern reported for developers trying to debug why anthropic sdk with mcp beta returns plain tool_use instead of expected mcp_tool_use, causing agent loops to hang. Based on the imported evidence, treat this as a tool-specific troubleshooting page rather than a generic API error.
Why this happens
GitHub Issue anthropics/anthropic-sdk-python#1581 (open, created 2026-05-21, 1 comment, 2605 related MCP/Auth issues in claude-code repo). Affects claude-sonnet-4-6 and other 4.x models under streaming API. Intermittent bug with multi-turn agentic loops mixing local function tools + mcp_toolset. This causes API billing charges for failed turns plus developer time wasted debugging. Category mapping: Anthropic SDK/MCP errors → Anthropic API per SKILL.md category table.
Common causes
- GitHub Issue anthropics/anthropic-sdk-python#1581 (open, created 2026-05-21, 1 comment, 2605 related MCP/Auth issues in claude-code repo). Affects claude-sonnet-4-6 and other 4.x models under streaming API. Intermittent bug with multi-turn agentic loops mixing local function tools + mcp_toolset. This causes API billing charges for failed turns plus developer time wasted debugging. Category mapping: Anthropic SDK/MCP errors → Anthropic API per SKILL.md category table.
Quick fixes
- Confirm the exact error signature matches
MCP connector model intermittently emits tool_use instead of mcp_tool_use for tools declared in mcp_toolset — conversation stalls on client side. - Check the Anthropic API account, local tool state, and provider configuration involved in the failing workflow.
- Compare the failing environment with a known working setup, then change one configuration value at a time.
Platform/tool-specific checks
- Verify the command, editor, extension, or API client that produced the error.
- Compare local settings with CI, deployment, or editor-level settings when the error appears in only one environment.
- Avoid deleting credentials, local model data, or project settings until the failing scope is clear.
Step-by-step troubleshooting
- Capture the exact error message and the command, editor action, or request that triggered it.
- Check whether the failure is account/auth, quota/rate, model/provider, local runtime, or deployment configuration.
- Review the source evidence below and compare it with your environment.
- Apply one change at a time and rerun the smallest failing action.
- Keep the working fix documented for the team or deployment environment.
How to prevent it
- Keep provider/tool configuration documented.
- Record non-secret diagnostics such as tool version, provider name, model name, and command path.
- Add a lightweight check before CI or production workflows depend on the tool.