AI & Machine Learning24 min
OpenAI Agents SDK maxTurns After Approval, Tested
I ran @openai/agents 0.19.0 and 0.20.0 side by side. In 0.20.0 the model call after an approval counts toward maxTurns, so a resume at maxTurns 1 can now throw.
Tag
#human-in-the-loop
5 posts
Security & Privacy21 min
ADK ToolCallIntegrityPlugin: Tamper Test Results
I rewrote an approved tool call in Google ADK TypeScript 2.2.1. It ran with no plugin and was blocked by ToolCallIntegrityPlugin, except in one setting.
AI & Machine Learning13 min
OpenAI Agents SDK needsApproval: Null Bug Tested
I ran OpenAI Agents JS 0.18.0 and 0.19.0 side by side. needsApproval got null for omitted optional fields until 0.19.0, and resumed runs now re-check it.
AI & Machine Learning35 min
Google ADK 2.0 Workflow: Retries and Timeouts Tested (2026)
ADK 2.0 Workflow retries are jittered, unmatched routes raise no error, and timeouts can't stop blocking code. Tested: google-adk 2.10.0, Python 3.10-3.13.
AI & Machine Learning13 min
Add Human-in-the-Loop Approval to Claude Agent SDK (2026)
Build a human-in-the-loop approval gate for the Claude Agent SDK in TypeScript: canUseTool prompts, PreToolUse hooks, and a full audit log for every tool call.
