Security Policy
How to report a vulnerability to Seele Agency, what to expect, and what we'll do (and won't do) once we hear from you.
Contact
Encrypt sensitive reports. If you can't encrypt, mention "security disclosure" in the subject line so it routes past spam filters.
What we promise
- Acknowledge within 48 hours of receipt.
- Triage within 7 days — confirmed, repro steps, and severity assigned.
- Fix or disclose timeline within 30 days of triage.
- Coordinated disclosure: 90 days from receipt, or until we mark the issue "fixed" — whichever comes first. We won't publicly disclose before then either.
- No legal action against researchers who follow this policy in good faith.
- Public acknowledgment on /security/acknowledgments (with your consent).
Scope
In scope
- seele.agency and any subdomain we operate (
www., docs., hub.).
- xkg-desktop — the desktop and CLI builds distributed from /downloads.
- xkg-server — the local sync server users self-host.
- cluster-hub — the cluster orchestration API on port 8090 (private; access via SSH/tailnet).
- xkg-mobile-flutter — TestFlight builds.
Out of scope
- Third-party AI providers — grok.com, claude.ai, chatgpt.com, gemini.google.com. Report to the vendor.
- Denial-of-service via volume.
- Physical attacks, social engineering, phishing of our staff.
- Automated vulnerability scanners without a specific target reproduction.
- Issues only present in outdated browsers (we support evergreen Chromium, Firefox, Safari).
Severity classification
- Critical — RCE, persistent auth bypass, full account takeover. Bounty: $500.
- High — Significant data exposure, privilege escalation (e.g. free Pro). Bounty: $200.
- Medium — Cross-origin data leakage, DoS affecting other users. Bounty: $75.
- Low — Information disclosure not compromising others. Bounty: $25.
- Info — Non-exploitable but useful hardening. Bounty: $10.
Payouts come from the bug-bounty pool, funded from 3% of net revenue. The pool is $0 today (no revenue yet). When revenue materializes, payouts flow to researchers who report valid issues via security.txt. See /security/acknowledgments for the pool balance, and /docs/bounty/ for the full program policy.
Program status: designed, not yet activated. The policy, payout tiers, response SLAs and safe-harbor terms are published in full at /docs/bounty/. The funding is pending: the revenue source has no live payment keys yet, so there is no revenue, no 3% allocation and no pool balance. Reports are still accepted and still fixed — only the payout is pending.
Safe harbor
When you conduct security research and report it to us per this policy, we will:
- Consider it authorized access under the CFAA, Computer Misuse Act, and equivalent laws.
- Not bring or support legal action against you.
- Not refer you to law enforcement for actions consistent with this policy.
- Work with you if a third party (e.g. your ISP) raises concerns, to clarify you were authorized.
What we won't do
- Threaten legal action against good-faith research.
- Retroactively redefine your report as "out of scope" after you've already written it up.
- Hide acknowledgment. Every valid report appears on /security/acknowledgments within 7 days of triage.
- Ask you to keep a critical vulnerability quiet indefinitely.
Acknowledgments
Researchers who report valid issues (with consent) are listed on /security/acknowledgments. The bug-bounty pool accrues at 3% of net revenue and pays out per the severity table above when revenue exists.
Safe harbor does not depend on being paid: a report is covered even if it is out of scope, a duplicate, rejected as invalid, or the pool is empty. Full safe-harbor terms, scope, exclusions and the activation checklist are in /docs/bounty/.