Public beta: v0.1.0-beta.1 — what I want tested

[ademisler]

RunOnMine v0.1.0-beta.1 is available for public testing on macOS, Windows and Linux.

The main design question behind RunOnMine is simple:

When an AI agent can touch a real machine, where should automation end and owner authority begin?

RunOnMine keeps that decision local. Requests are evaluated using the connector/requester, tool, targeted resource and local policy, then allowed, denied or held for local approval.

What I would like tested

• installation and first-run onboarding
• selected-root filesystem behavior
• Safe / Developer / Automation policy profiles
• approval wording and granularity
• connector setup and recovery
• browser and desktop workflows
• Emergency Lock and reconnect behavior
• diagnostics and troubleshooting
• threat-model assumptions or missing boundaries

Beta limitations

RunOnMine is a security boundary and approval system, not a sandbox. A permitted shell, browser or desktop action can create real side effects with the authority of the operating-system account running it.

The current beta also has distribution-trust gaps that are intentionally documented:

• macOS is ad-hoc signed, not Developer ID signed/notarized
• Windows is not Authenticode signed
• an independent external security review is still pending

Release artifacts include SHA-256 checksums and CycloneDX SBOMs. The repository also contains the threat model and machine-readable release/acceptance evidence.

Links

Repository: https://github.com/ademisler/RunOnMine

Release: https://github.com/ademisler/RunOnMine/releases/tag/v0.1.0-beta.1

Documentation: https://github.com/ademisler/RunOnMine/tree/main/docs

For security vulnerabilities, please use GitHub private vulnerability reporting rather than posting exploit details publicly.

For everything else, bug reports, permission-model criticism and real workflow examples are especially useful at this stage.