PreToolUse
PulseAugur coverage of PreToolUse — every cluster mentioning PreToolUse across labs, papers, and developer communities, ranked by signal.
10 day(s) with sentiment data
-
Claude Code enhanced with security hooks to block dangerous commands
Developers using Claude Code can enhance its safety by implementing shell hooks that automatically block dangerous commands and prevent access to sensitive files. These hooks, executed before or after specific tool uses…
-
Claude Code agent bypasses URL-opening hook, opens pages in wrong browser profile
A developer encountered an issue with their Claude Code agent on macOS where the `open` command, intended to open URLs in a specific browser profile, failed to do so correctly. The agent's `PreToolUse` hook, designed to…
-
Claude Code hooks fail to block commands on exit code 1
A user discovered a critical flaw in Claude Code's hook system where exit code 1, commonly understood as an error in Unix-like systems, does not actually block tool execution. Instead, only exit code 2 is recognized as …
-
Claude Code bypasses found; new hook blocks force pushes
A security researcher tested 22 methods to bypass Claude Code's rules for preventing force pushes to main branches, finding that 15 of the methods initially succeeded. While Claude Code's built-in deny rules blocked 7 a…
-
AI code reviewer repeatedly used same model as writer, violating rule
A developer discovered that their AI agent, designed to review code written by another AI model, was frequently using the same model for both tasks, violating a core rule. This oversight meant that 89 out of 190 reviewe…
-
Claude Code adds security layer via settings.json for shell command permissions
Claude Code now allows users to define shell command permissions in a settings.json file, rather than solely relying on prompts. This approach provides a more robust security measure by enabling an allow/deny list that …
-
Claude Code hook matching fails on exact server prefix, documentation mismatch found
A technical deep-dive reveals a significant discrepancy in how Claude Code handles tool matching for hooks. While documentation and permission rules suggest that a server prefix like 'mcp__lab' should match all tools fr…
-
Developer creates safety hook to prevent Claude Code from destructive commands
A developer has created a 15-line Python script designed to act as a safety hook for Claude Code, a coding agent. This hook, named PreToolUse, intercepts potentially destructive commands before they are executed by the …
-
Developer shares strategies for managing multiple Claude Code sessions on a single repo
A developer is exploring methods to manage multiple instances of Claude Code operating on the same repository checkout without using separate worktrees. The current approach involves a set of rules and a small tool to p…
-
AI agent details common CLAUDE.md configuration mistakes
This article details common mistakes made when configuring CLAUDE.md files for the Claude Code agent, emphasizing that these files are context rather than configuration. The author, an AI agent named Claude, highlights …
-
Claude Code Hooks Automate Formatting and Block Dangerous Commands
Developers can now implement custom hooks for Claude Code to automate tasks and enforce safety rules. These hooks, configured in `.claude/settings.json`, can automatically format code after edits or block dangerous comm…
-
Claude Code hooks fail on Windows, exposing users to risks
A technical post details how Claude Code's safety hooks can fail on Windows, potentially allowing unintended actions. The author outlines nine specific ways these hooks can malfunction, often due to differences in shell…
-
Developer creates Claude Hookbook for automated Claude Code tasks
A developer has created "Claude Hookbook," a library of pre-written hooks designed to automate common tasks and add guardrails to Anthropic's Claude Code. The tool addresses the setup friction of manually configuring ho…
-
Claude Code enhances AI agent security with multi-layer permissions
Claude Code offers a robust permission system for AI agents, moving beyond simple yes/no prompts. It utilizes a three-layer architecture involving interactive approvals, static deny/allow lists in settings.json, and dyn…
-
Claude Code hooks fail silently due to unobservable errors
This article details several ways Claude Code hooks can fail without producing errors, leading to silent malfunctions. Issues include event types that ignore matchers, 'if' conditions only evaluating on tool events, dep…
-
Developer shares method to measure Claude Code hook latency
A developer has detailed a method for measuring the latency of custom scripts, known as "hooks," integrated into Claude Code. These hooks, used for tasks like automatic git commits or project categorization, can signifi…
-
Claude Code hook bug halts tool calls after first denial
A user encountered an issue with Claude Code where tool-use hooks stopped functioning after the first denial. The user attempted to reduce token usage by redirecting file reads to a local LLM summarization tool via a Pr…
-
Developer shares Claude Code usage tips and fixes for common mistakes
A developer shares their experiences and lessons learned from using Claude Code as their primary coding assistant for over a year. They detail six common mistakes, including repeating instructions instead of using CLAUD…
-
Claude Code hooks: troubleshooting guide for custom script execution
This guide explains how to troubleshoot issues with Claude Code hooks, which are custom scripts designed to automate actions or enforce policies within the Claude Code environment. It details steps to verify if a hook i…
-
Claude Code v2.1.211 fixes silent override of Bash execution guardrails
Claude Code version v2.1.211 has fixed a critical bug where its auto mode would silently override user-configured PreToolUse hooks designed to prompt before executing unsandboxed Bash commands. Previously, if a hook ret…