Skip to content

Security: ObtuseAI/nimrod

SECURITY.md

Security policy

nimrod is a defensive research preview. It is not safe to deploy as a production protection or response system.

Reporting a vulnerability

Use this repository's Security tab and Report a vulnerability to open a private GitHub security advisory. Do not disclose suspected vulnerabilities, authorization bypasses, target-widening paths, unsafe connectors, live credentials, or exploit details in a public issue.

Include the affected revision, component, preconditions, source-to-sink path, impact, and a minimal non-destructive reproducer. Good-faith, authorized research that respects the license, avoids privacy violations and service disruption, and gives the maintainers reasonable remediation time is welcome.

Supported versions

Only the current default branch and latest tagged public research preview are supported. Replayed fixtures, historical private-archive commits, range packets, and generated reports are not independently supported products.

Security claims

Only claims mapped to reproducible release evidence may be made. Passing tests, valid signatures, model judgment, successful API responses, and completed builds do not by themselves prove prevention, detection, containment, response, recovery, or production safety.

Prohibited behavior

nimrod may not be used for unauthorized access, counter-hacking, malware, surveillance, destructive remediation, credential collection, target enumeration outside explicit scope, or critical-infrastructure control.

Release hygiene

Public releases require clean secret and private-path scans, dependency review, canonical manifest and contract validation, explicit origin labeling, read-only workflow permissions, and exact-SHA quality proof.

There aren't any published security advisories