OASM is pre-1.0 (0.0.1 in package.json). Only the latest main branch and the most recent release receive security fixes. There is no LTS series.

Use GitHub's private vulnerability reporting — open a draft Security Advisory. This keeps details private until a fix is ready. There is no dedicated security email address; do not post exploit details in public issues or Discord.

If the issue is clearly not sensitive (hardening idea, dependency bump, theoretical concern with no exploit path), a public bug report is fine.

• Affected component (core-api, console, worker, oasm-connectors manifest) and version/commit.
• Steps to reproduce or proof of concept.
• Impact assessment: what an attacker gains (data access, cross-tenant access, RCE, auth bypass).
• Your environment if relevant (Docker Compose, dev mode, production build).

• Reports are triaged by maintainers via the private advisory thread.
• There is no formal SLA — response time depends on severity and maintainer availability.
• Once triaged, fixes land on main and ship with the next release; reporters are credited unless they opt out.

• Worker Docker socket is by design. The worker mounts /var/run/docker.sock to spawn connector containers (docker-compose.yml, worker/internal/runtime/docker.go). The socket is root-equivalent on the host — only run trusted connector images, never expose the worker to untrusted networks.
• Secrets live in .env, never in git. BETTER_AUTH_SECRET signs sessions and email verification links, and the last ENCRYPTION_KEYS entry encrypts stored credentials; outside local development the API refuses to start unless both are random 32+ character values. WORKER_API_KEY must match what the API expects. Leaked-key rotation is an operator responsibility.
• .env / deployment misconfiguration is out of code scope for reports — flag it as a docs issue instead, unless the defect is in source code (e.g. an insecure default constant).