OpenMoat: open-source local policy checks to secure AI coding agents with machine access
__dotk__ · reddit · 2026-11-01
A Reddit user explores the security risks of AI coding agents (Claude Code, Codex, Cursor) that can read files, run commands, install packages, and reach external services — especially when an agent follows malicious instructions embedded in untrusted files or tool responses, which permission prompts alone don't cover.
They introduce OpenMoat, an open-source project that checks agent actions against local security policies (github.com/crocodile-labs/openmoat), and asks the community:
- Do you run coding agents in a sandbox?
- How do you protect credentials and sensitive files?
- How do you balance security with developer productivity?
More from coding & agent
- Indie dev launches EasyComfy Pro: ComfyUI workflows as one-click presets with MCP support — Dependent-Heat3770 · 2026-11-01
- Human-in-the-Loop theater: Devs rubber-stamp Claude's 400-line 'dangerous' bash scripts — rickasaurus · 2026-11-01
- One Claude Code command audits your Managed Agents setup against Anthropic's security guide — brada · 2026-11-01
- Claude-Code-Game-Studios: open-source template turns Claude Code into a 49-agent game studio — lxfater · 2026-11-01
- Senior engineer: 'popularity is the weights of an LLM' — half of 600 engineering scars can't be baked into models — RandalSchwartz · 2026-11-01
- Factory's agent ships an MCP connector registry; david hoang wants Paper added — davidhoang · 2026-11-01