From 973296210fa9e29c52d2d756fcf299e7c59410d5 Mon Sep 17 00:00:00 2001 From: yuhengcai1 Date: Mon, 26 Oct 2026 00:04:51 -0700 Subject: [PATCH] docs(sdk): clarify comparison of invalid tool calls Co-authored-by: openhands --- sdk/guides/agent-stuck-detector.mdx | 1 + 1 file changed, 1 insertion(+) diff --git a/sdk/guides/agent-stuck-detector.mdx b/sdk/guides/agent-stuck-detector.mdx index 06846dc28..85d281e41 100644 --- a/sdk/guides/agent-stuck-detector.mdx +++ b/sdk/guides/agent-stuck-detector.mdx @@ -37,6 +37,7 @@ stuck status at any point using `conversation.stuck_detector.is_stuck()`. The stuck detector compares events based on their semantic content rather than object identity. For example: - **Actions** are compared by their tool name, action content, and thought (ignoring IDs and metrics) + - When a tool call fails validation and has no executable action, the detector also compares the retained argument strings. Different invalid arguments do not count as the same call merely because both actions are absent. - **Observations** are compared by their observation content and tool name - **Errors** are compared by their error messages - **Messages** are compared by their content and source