Management OS, Inc. welcomes reports from security researchers. This file summarizes our published vulnerability disclosure policy. If the two ever differ, the published policy governs.

Report privately, in either of two ways:

• On GitHub: open the affected repository's Security tab and choose Report a vulnerability. These reports are handled under the same published policy.
• By email: security@mgmt-os.ai. If the report is sensitive, ask for a secure channel and we will arrange one.

Include a description of the issue, the steps to reproduce it, and any proof of concept that helps us confirm it. Please don't open a public issue for a security problem.

Machine-readable contact: mgmt-os.ai/.well-known/security.txt

• We acknowledge every report, typically within two business days.
• We investigate promptly, keep you updated on our assessment, and tell you when the issue is resolved.
• We credit you for the discovery if you'd like, once the issue is fixed.

We don't run a paid bug bounty. We offer public acknowledgment.

• Give us a reasonable window to investigate and fix the issue before any public disclosure.
• Make a good-faith effort to avoid privacy violations, data destruction and any interruption to the service.
• Only use accounts you own or have explicit permission to test. Do not access, change or copy other customers' data.
• Don't run automated scanning that degrades the service, and don't use social engineering, physical attacks or denial of service.

We will not pursue or support legal action against researchers who act in good faith and in accordance with the published policy. The full safe-harbor terms are in the policy.

In scope: the Management OS application and its APIs at mgmt-os.ai and its subdomains. That includes the MCP server at https://app.mgmt-os.ai/api/mcp, which the plugin and connectors in this organization use.

For the AI-assistant integration, we especially want to hear about anything that:

• lets one workspace read another workspace's data
• gets past sign-in, or uses a token for something it wasn't granted
• makes a read-only tool send, change or delete anything
• lets text inside shared evidence steer a tool into revealing data it shouldn't

Out of scope: our sub-processors' own infrastructure (report those to the provider directly), findings that require a compromised device or account, and best-practice suggestions without a demonstrable security impact. Our sub-processors are listed at mgmt-os.ai/subprocessors.

The management-os-plugin repository holds no code and no credentials; its tools run on Management OS servers. If you find a file in any of our public repositories that points somewhere it shouldn't, report it the same way.