HOL Guard Open Source: A Local Security Layer for AI Coding Agents
kantorcodes1 · reddit · 2026-08-13
A development team has released the open-source tool HOL Guard, designed to add a local security policy layer for AI coding agents. It is compatible with popular tools like Gemini CLI, Codex, Claude Code, and Cursor.
Core mechanics:
- Establishes an interception layer between the AI agent and the local machine, checking commands, tool calls, MCP server interactions, package installations, and file access before execution.
- Supports allowing, warning, asking, or blocking actions based on user-defined policies.
- Logs every decision receipt so developers can trace exactly what the agent did afterward.
The tool is completely free, open-source, and runs locally without requiring a cloud account.
More from coding & agent
- Claude Transforms Napkin Sketch into CAD Model in Under 10 Minutes — burhop · 2026-08-13
- Claude Agent Turns Napkin Sketch into CAD Model in 10 Minutes — floguo · 2026-08-13
- Paper Proposes 'Sleep Phase' for LLMs to Consolidate Memory Offline — rohanpaul_ai · 2026-08-13
- Mine Agent Transcripts for Correction Patterns to Improve Evaluations — jonah_omninode · 2026-08-13
- Correcting an AI Agent Twice Means a Missing Workflow Rule — jonah_omninode · 2026-08-13
- Dopamine Loop for AI Coding Agents: Open-Source Skill Cuts Costs by 30% — AutoProspectAI · 2026-08-13