gateguard
affaan-m/everything-claude-code
GateGuard acts as a powerful pre-tool-use hook, forcing the LLM to undergo deep, factual investigation before allowing any code edits, file writes, or destructive Bash commands. Unlike simple self-assessment, it requires citing specific facts (e.g., imports, data schemas, affected functions), significantly increasing the reliability and depth of AI-generated outputs. Ideal for complex coding tasks, data manipulation, and high-stakes development environments.