Skip to content

Security: OpenSIN-AI/OpenSIN-overview

Security

SECURITY.md

Security Policy

Scope: This document covers the OpenSIN-AI/OpenSIN-overview governance repo. For vulnerabilities in the runtime engine, CLI, backend, WebApp, or any other OpenSIN-AI repo, open the report against that repo.

Why a governance repo needs a security policy: the scripts, schemas, and registries here feed marketplace frontends and CI pipelines across ~200 downstream repos. A compromised team.json, a poisoned oh-my-sin.json, or a tampered MASTER_INDEX.md can cascade into every surface that renders it. Treat this repo accordingly.


1. Supported versions

Version Supported
main branch (HEAD)
Any tagged release ≤ 30 days old
Anything older ❌ — fix must be rebased onto main

This repo is a rolling governance index, not a versioned library. There are no long-lived stable branches.


2. Reporting a vulnerability

Do NOT open a public GitHub issue.

Use one of these private channels in priority order:

  1. Preferred — GitHub Private Vulnerability Reporting (docs) This creates a private advisory visible only to maintainers. First-class tooling, no attachments lost, full audit trail.

  2. Fallback — Email: security@opensin.ai PGP optional (key fingerprint published in a pinned issue on this repo once available).

  3. Emergency during launch week (2026-04-19 → 2026-04-30): ping the on-call Discord role @launch-oncall in the private #security channel. Use this only for SEV-1 (actively exploited, data loss, RCE).

Please include:

  • A clear description of the issue and the repo/file/line it affects.
  • Reproduction steps, or a proof-of-concept, or a pointer to the commit/PR introducing the issue.
  • Your assessment of severity (SEV-1 / SEV-2 / SEV-3) using the incident template's severity guide.
  • Whether you are requesting a CVE assignment.

3. Response SLAs

Severity Acknowledgement Triage Fix or mitigation
SEV-1 (active exploit / RCE / data loss) 4 h 24 h 48 h
SEV-2 (privilege escalation / authn bypass) 24 h 72 h 7 d
SEV-3 (info leak / tampering / DoS) 72 h 14 d 30 d

SLAs are measured in wall-clock time during launch week (2026-04-192026-04-30) and in business hours (CET) outside of it.


4. Disclosure policy

We follow coordinated disclosure.

  1. You report privately (§2).
  2. We acknowledge (§3) and triage.
  3. We confirm the fix in a private PR and credit you (if desired) in the release notes.
  4. We publish a GitHub Security Advisory after the fix lands on main and downstream consumers have had a reasonable window to update.
  5. 90-day maximum embargo. If we cannot fix a SEV-1/SEV-2 within 90 days, we publish the advisory anyway and ship whatever mitigation we have — silence is not a strategy.

You are welcome to publish independently after the advisory is public, or earlier with written agreement.


5. What counts as a vulnerability in this repo

Because this repo is a governance index, "vulnerability" has a narrower meaning than in a product repo. In-scope:

  • Registry tampering — changes to registry/MASTER_INDEX.md, registry/SCAFFOLD_AUDIT.md, templates/oh-my-sin.json, platforms/canonical-repos.json, or any templates/teams/*.json that are not produced by the documented pipeline and go unreviewed.
  • Script RCE — any file under scripts/ that executes untrusted input (filename, issue body, API response) without sanitisation.
  • Workflow privilege escalation — any GitHub Actions workflow (in .github/workflows/ or governance/workflows-proposed/) that leaks secrets, runs untrusted PR code with write tokens, or bypasses CODEOWNERS.
  • Schema escape — any schemas/*.schema.json that allows values that downstream consumers (marketplace UI, chat.opensin.ai dashboard) interpret as HTML/JS/SQL without further validation. Schemas here are the first line of defence for the marketplace.
  • Typosquatting / dependency confusion — since package.json is private: true and we run zero runtime dependencies, this is extremely unlikely — but a PR adding an npm dependency that resolves to a malicious registry package is in scope.
  • Chain of trust breaks — anything that lets a non-CODEOWNER merge into main without review.

Out of scope in this repo (report in the canonical repo instead):

  • Runtime vulnerabilities in OpenSIN (Python engine) → file with OpenSIN-AI/OpenSIN.
  • Vulnerabilities in OpenSIN-Code (CLI), OpenSIN-backend (control plane), OpenSIN-WebApp (chat.opensin.ai), or the websites → file with that specific repo.
  • Issues with individual A2A-SIN-* agents → file with that agent repo.
  • OpenCode config issues → OpenSIN-AI/Infra-SIN-OpenCode-Stack.
  • PCPM / memory daemon → OpenSIN-AI/Infra-SIN-Global-Brain.

See BOUNDARIES.md for the full canonical-ownership map.


6. Hall of fame

We credit reporters in the release notes and in a docs/security/hall-of-fame.md file (created on first report). If you prefer to remain anonymous, say so in the report.


7. Dependencies and automated scanning

This repo pins zero runtime dependencies on purpose (package.json declares private: true, no dependencies / devDependencies at the time of writing). The only recurring supply-chain surface is GitHub Actions pin updates. Those are handled automatically by Dependabot.

If you spot an unpinned action reference, an outdated major, or a dependency that snuck into package.json, open a PR — that itself counts as a SEV-3 report.


Last reviewed: 2026-04-19.

There aren’t any published security advisories