TypeS is a reference implementation of an architecture, not a battle-tested product with a release history. Before relying on it for anything sensitive, read docs/PRODUCTION-READINESS.md: it is a candid, ranked list of what stands between this codebase and real production traffic — a hardened policy engine, a threat model at the boundary, data-classification enforcement, encryption in transit and at rest, secrets management, and more.

Items already documented there are known limitations, not vulnerabilities. Please don't file a report telling us there is no at-rest encryption or that the bundled AbacPolicyEngine hasn't had a security review — we know, it's written down, and a report restates the gap rather than revealing one. A report is valuable when it shows behaviour that contradicts what the docs claim, or a weakness the docs don't already name.

Every @typesys/* package is pre-1.0 (0.x) and nothing has been published to a registry (ADR-0020). There are no maintained release branches: security fixes land on main. If you have vendored or forked the code, track main for fixes.

Please report privately — do not open a public issue for a suspected vulnerability.

Use GitHub's private vulnerability reporting: on the repository's Security tab, choose "Report a vulnerability". This opens a private advisory visible only to you and the maintainers.

Please include, as far as you can:

• the affected package(s) and file(s), and the commit SHA you observed it on;
• a description of the issue and its impact (what an attacker gains);
• a minimal reproduction — a failing test, a curl against the HTTP transport, or a short script is ideal;
• any suggested remediation.

This is a small project, so response is best-effort rather than SLA-backed. We aim to acknowledge a report within a week, agree on severity and a fix approach with you, and credit you in the advisory and the changeset unless you'd rather remain anonymous. Please give us a reasonable chance to ship a fix before disclosing publicly.

In scope: anything that lets a caller cross the one governed boundary the architecture promises — reading an object, property, or relationship a policy should deny; invoking an Action without passing its policy check; escaping the query/Action input bounds (InputValidator); leaking a value through provenance, an error message, a query filter/sort, or an aggregation that policy should have redacted; or breaking the identity resolution in @typesys/auth-oidc.

Out of scope: the documented production-readiness gaps (docs/PRODUCTION-READINESS.md), issues in the example domain-airforce / domain-hospital domains that exist only to demonstrate the architecture, and anything requiring a threat model or deployment hardening the project explicitly hasn't done yet.