Map / AI Endpoint Security / HOL Guard
HOL Guard
hol.org/guard ↗ · Open on the map
AI Endpoint SecurityRuntime & GuardrailsMCP & LLM Gateways
Capabilities
Open-source (Apache-2.0) local security layer for AI coding agents. It reviews shell commands, file access, package installs, MCP tool calls, plugins and skills before they run, and keeps a searchable local trail of every decision. An optional paid Guard Cloud adds synced org-wide policies, telemetry and team dashboards.
Sensors / integration
Runs on the developer's machine and attaches to agents through native pre-tool hooks, managed MCP proxies and launch integrations. Supports Claude Code, Codex, Cursor, GitHub Copilot CLI, Gemini CLI, Cline, OpenCode, Antigravity, Devin and others. Works offline with no cloud dependency.
Protections
Applies allow, observe, ask or block decisions to destructive file and shell operations, credential and secret access, prompt-injection-driven commands, suspicious outbound transfers, malicious packages, unsafe MCP server configs, risky plugin or skill settings, and unsafe git pushes to production branches.
Differentiator
One local policy across many coding agents, enforced on-device before the action executes. Built by HOL (Hashgraph Online), a DAO-structured standards body for AI agents on Hedera, rather than a venture-backed security vendor.
Team & funding
- Founders
- Michael Kantor
Other AI Endpoint Security companies
Anomity · AstraSync AI · Asymptote Labs · Autonomous · Backslash · Bay Security · Bloom · BlueRock · Bold · Canyon Road · Capsule Security · Certiv · See all →
Last updated · Get the “Listed on” badge