ACCESS_CODE unset remains fail-open in middleware, document reads are
capability-by-id via the anonymous owner cookie (not x-learner-key),
and the README action/skill counts match the Action union and
skills/agent-runtime.
Closes#1587
Co-authored-by: wyuc <wang-yc24@mails.tsinghua.edu.cn>
Add a "Before You Report" section (reproduce on the latest release, check
published advisories, prefer the default deployment) and a "Deployment
Assumptions" section describing the boundaries reports are assessed against:
ACCESS_CODE is a shared password, the render service isolates only with its
shipped egress lockdown, forwarding headers are trusted only with
TRUST_PROXY_HEADERS, PERSISTENCE_DEV_TOKEN is not user isolation, and
operator-set endpoints are trusted configuration.
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
Add a triage section that says how severity is assigned and that
severity objections are answered in the advisory thread before
publication, and a disclosure note that maintainers request the CVE
through GitHub at publication time and keep the published advisory
consistent with the CVE record.
Claude-Session: https://claude.ai/code/session_01GRaNc5E88r41GUWt2Y3uiQ
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>