Skip to content

Security: agentscope-ai/QwenPaw-Data

Security

SECURITY.md

Security Policy

Supported Versions

QwenPaw-Data is pre-1.0 software. Only the latest release on the default branch receives security fixes.

Deployment Model

QwenPaw-Data is designed for local-first, single-user deployments. All services bind to 127.0.0.1 by default. Before exposing any service beyond loopback, read the "Security Model and Known Limitations" section in README.md and configure QWENPAW_DATA_API_TOKEN or scoped QWENPAW_DATA_API_KEYS, plus an exact QWENPAW_DATA_CORS_ORIGINS origin allowlist.

Known, intentionally documented limitations:

  • The Host defaults to a per-session Docker workspace. The explicit workspace_type="local" / --workspace local escape hatch executes agent shell commands on the host without a sandbox. Docker resource limits, egress network policy and non-root execution are not yet applied (roadmap), so the container boundary is not a hardened multi-tenant sandbox.
  • Scoped API keys provide endpoint authorization but not a multi-user identity provider or user lifecycle/RBAC system.
  • Authentication-failure and request-rate limiter state is process-local; multi-worker or horizontally scaled deployments require a shared limiter.
  • Outbound import callbacks are disabled unless their exact origin is listed in QWENPAW_DATA_CALLBACK_ALLOWLIST; DNS answers are pinned per request and request/response sizes and total duration are bounded.

Reporting a Vulnerability

Please do not open a public GitHub issue for security problems.

Instead, report vulnerabilities privately via GitHub Security Advisories ("Report a vulnerability" on the repository's Security tab). If that is not available to you, contact the maintainers listed in the repository profile.

Please include:

  • A description of the issue and its impact.
  • Steps to reproduce (proof-of-concept if possible).
  • Affected version/commit and environment details.

We aim to acknowledge reports within 7 days and to release a fix or mitigation guidance within 90 days of confirmation. We will credit reporters in release notes unless you request otherwise.

There aren't any published security advisories