Risky agent settings
guard scan reports bypass and never-ask modes, broad allow rules and disabled hooks on a machine, locally and with nothing uploaded.
Free local assessment →Who it’s for
Ship with coding agents, not around them.
Your developers already run Claude Code, Codex, Cursor and more, plus MCP servers and agents in CI. Runeward Labs lets them keep that speed while security sees what the agents do.
What worries you
And what Runewrd Labs does about it.
MCP servers you didn’t vet
Guard pins MCP tool definitions and reports changes, and flags servers that aren’t on your approved list.
MCP supply-chain hygiene →Agents in CI and automation
Rune Box runs agents in governed environments: allow, deny or require approval before execution, with deny-by-default egress.
Rune Box →Deploying it
How it fits
- Startup: Guard hosted by Runeward Labs for your team, per user.
- Rune Box is free and open source under Apache-2.0.
Honest limitations
What to know first
- Runeward Labs holds no security certifications. Framework mappings in findings are candidate mappings for triage, not compliance evidence.
- Rune Guard is in early access, and what it covers differs by platform and harness: see the known limits before relying on a signal.
Your agents already have access. Ward off the risks.
Join Guard early access for your team, or run the free local scan today. Agents in automation too? Add the open-source Rune Box.