Coordinated Vulnerability Disclosure Policy Whisper Security · viaGraph b.v. · the Whisper platform Last updated: 2026-09-17 We welcome good-faith security research on the platform we operate, and we will work with you to verify, fix, and disclose issues responsibly. This policy covers the Whisper PLATFORM. The network layer we operate (AS219419, our routing, and the prefixes we originate) has its own policy, with its own scope and contacts: https://as219419.net/.well-known/disclosure-policy.txt Both reach the same mailbox, so send your report wherever it fits best. If you are unsure, just send it; we will route it. How to report ------------- Email: security@whisper.security Encrypt: please encrypt sensitive reports with our OpenPGP key, published at https://whisper.online/.well-known/openpgp-key.txt fingerprint EFF1 663D 9925 3968 2106 A5EA D0F7 0908 CF3B 7929 Include: a clear description, the affected surface or endpoint, reproduction steps or a proof-of-concept, and the impact you observed. One issue per report is easiest to track. In scope -------- - whisper.online and its subdomains, the content they serve, and the public agent-facing surfaces on them: the discovery documents, the onboarding manifest, the Agent Card and the A2A endpoint it advertises. - The control plane and the graph API, including the agent verbs reached through them and the authorization that gates each one. - The identity plane as infrastructure: how agent /128s are allocated, registered, resolved, reverse-resolved, and published over RDAP, and the certificate and DNSSEC machinery behind them. - The Whisper CLI, the SDKs and the MCP server we publish, and the release and signing path that delivers them. A supply-chain finding against an artifact we sign is firmly in scope. - The endpoint sensor and the enrollment it performs. - The console and the authentication in front of it. Out of scope ------------ - The individual agent /128 hosts in 2a04:2a01::/32 and any data, traffic, or conduct on them. Do not access, probe, or interfere with them. (The identity infrastructure above is in scope; the agents themselves are not.) - Customer tenants and their data. Test against your own account only. - Anything we do not operate: third-party networks and platforms, our providers, and destinations reached through our egress. - Denial-of-service, volumetric, or stress testing, and anything that degrades service for others. Do not run these. - Social engineering or phishing of our people, and physical attacks. - Findings with no security impact (missing headers on static no-auth content, version banners, intentionally-open ports, and similar). Safe harbour ------------ If you make a good-faith effort to follow this policy, we will not pursue or support legal action against you for your research. Good faith means: stay in scope; access only the data needed to demonstrate the issue and never store, modify, or share data that is not yours; do not degrade or disrupt our or others' services; give us reasonable time to remediate before any public disclosure; and stop and contact us if you encounter personal data. Our commitment -------------- - We acknowledge reports as quickly as we can, on a best-effort basis during RIPE-region business hours. We are not a 24x7 NOC. - We will keep you updated through validation and remediation, and we will agree a disclosure timeline with you: typically up to 90 days, sooner for low-risk issues, longer by mutual agreement where a fix is complex. - With your consent we are glad to credit you (see Acknowledgments in .well-known/security.txt). Other contacts -------------- Abuse (spam, traffic, routing complaints): security@whisper.security (same mailbox as reports, by design; it is our RIPE abuse-c) Peering / interconnection: peering@whisper.security Operational / NOC: noc@whisper.security