开源本地防护工具Axiom发布，拦截AI编码智能体越界写入
Start here
2026年11月6日，开源本地护栏工具 Axiom 发布。该工具面向 AI 编码智能体，支持在 Claude Code 和 Cursor 等工具写入代码前实施预拦截。Axiom 不依赖外部 API、Token 或第三方依赖，通过正则与 AST 模式匹配提取约束，并基于三态逻辑执行阻断或确认。在防护范围上，该工具覆盖数值超限、硬编码鉴权绕过、高危 Bash 命令与密钥泄露等常见意外违规，但不支持对抗性混淆防御。
AI-generated from reports · Updated 5 hours ago
Report timeline
Follow the reports to see different sides of the story.
- Hacker News · AISelectedOpen-source local guardrail tool Axiom launches to intercept out-of-bounds writes by AI coding agents
Axiom is a local guardrail tool designed for AI coding agents, providing pre-write interception before tools like Claude Code and Cursor apply code changes. Requiring no external APIs, tokens, or third-party dependencies, it extracts constraints via regex and AST pattern matching and enforces blocks or user confirmations based on tri-state logic. The tool covers common accidental violations, such as out-of-range numerical values, hardcoded authentication bypasses, high-risk Bash commands, and credential leaks, though it does not defend against adversarial obfuscation.
Story heat trend
Current heat 9·Comparable peak 10（Nov 6 10:00）·Comparable change, last 24 hours–
Trends compare only the same parties observed continuously, so the scope can be narrower than current heat counts. Move the pointer or tap the chart to see hourly heat; use the left and right arrow keys on a keyboard.