A bug in Claude Code version 2.1.319 allowed commands to execute without permission prompts when multiple PreToolUse hooks were involved. Specifically, if a hook returning 'defer' finished before a hook returning 'ask', the command would run without the expected prompt. The subsequent version, 2.1.322, corrected this behavior, ensuring the prompt appeared as intended. This issue is significant for agent repositories that rely on these hooks for security and control, as the interaction between different hook types and their timing could bypass intended safeguards. AI
IMPACT This fix ensures that security prompts function correctly in Claude Code, preventing unintended command execution and maintaining the integrity of agent workflows.
RANK_REASON The item details a specific bug fix in a software product, Claude Code, related to its hook execution and permission prompting.
Read on dev.to — Claude Code tag →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →