Results / truncation.tool-call-args
Tool-call arguments cut off by the output token limit
The model stops at the output token limit partway through the arguments of a tool call.
Bug class: Tool calls lost, merged or returned with broken arguments
The library drops or merges tool calls, mixes up their ids, or returns a call as complete while its arguments are cut off or repaired.
Tools run with wrong or incomplete arguments, or don't run at all.
Requirements
| Level | Id | Check | Requirement | Providers |
|---|---|---|---|---|
| MUST | no-complete-partial-call | no_complete_tool_call_from_partial_args | Partial tool-call arguments are never presented as a complete parsed call. | All |
| MUST | cause-is-truncation | truncation_is_signalled | The caller can tell that the output stopped at the token limit. | All |
Fixtures
| Provider | Tier | any-llm | litellm | pydantic-ai | tanstack-ai | vercel-ai |
|---|---|---|---|---|---|---|
| anthropic | documented | ✓ PASS | ✓ PASS | ✓ PASS | – UNSUPPORTED | ✓ PASS |
| openai-chat | recorded | ✓ PASS | ✓ PASS | ~ LOSSY | – UNSUPPORTED | ✓ PASS |
| openai-responses | recorded | ✓ PASS | ✓ PASS | ~ LOSSY | – UNSUPPORTED | ✓ PASS |