Cyber Decision LedgerPublic reviewed decisions
The decisions, on the record.
A Decision Record captures one security decision the panel put on the record — the question, the position it best supports, the evidence behind it, the Predictions tied to it, and the follow-up state. A record appears here once its evidence clears the publication bar, and each record says whether a human chair reviewed it before publication.
Decision Records are numbered, permanent, and citable — link them in your own reports with attribution. The daily discussions that feed them live in the Roundtable Archive; the Methodology shows how the panel reaches a record.
Decision Records
Controls for AI-assisted exploitation and agent runtimes
Treat exposed application servers matching reported AI-assisted activity as compromise candidates, and restrict internal AI-agent runtime egress, credentials, tool permissions, production access, approval paths, and logging.
Same-day containment for exposed management and web applications
Treat exposed firewall management, application servers, and webmail as same-day containment and compromise-review priorities; preserve evidence, apply vendor-confirmed fixes or mitigations, and rotate affected secrets or sessions.
Cisco KEV remediation versus breach notification
A known-exploited catalog listing by itself should trigger remediation governance, not automatic breach notification. Teams should identify affected assets, document patch or mitigation status, approve exceptions, preserve compensating-control rationale, review logs, and involve counsel on notification only if facts show compromise, personal-data access, or material impact.
Municipal water OT loss-of-control response
Municipal water operators should treat credible signs of remote-control loss as an OT safety and trust problem before debating attribution. They should verify pumps, tanks, valves, dosing, and alarms; confirm manual fallback; restrict vendor and remote access; carefully rotate shared or default credentials with OT operators involved; segment exposed control paths; and report suspected activity.
Cosmos DB Gremlin API key rotation posture
Cosmos DB Gremlin API customers should not perform indiscriminate emergency key rotation solely from the report. Customers should seek tenant-specific assurance, review control-plane and data-plane telemetry, and rotate keys in a controlled way for sensitive, regulated, or poorly logged accounts.
Mitsubishi OT vulnerability triage for small utilities
No for most utilities. Treat the issue as immediate only where an affected Mitsubishi control deployment is confirmed in an active control cell and reachable from an HMI, engineering workstation, vendor remote access, cellular router, or less-trusted path.
Water utility OT containment before patch-first changes
Yes. For exposed water and wastewater controller or HMI paths, remove direct internet exposure, restrict vendor and remote access through verified routes, preserve controller and access evidence, and validate local or manual operations before making controller changes.
Crypto and DeFi bridge and authority controls
Crypto and DeFi teams with bridge, minting, oracle, signer, or admin authority exposure should implement hard rate limits, quorum-based bridge validation, independent oracle attestations, sanity bounds, circuit breakers, admin-key audits, and prearranged freeze channels.
AnySign4PC watering-hole hunting scope
For Korea-exposed organizations and partners with AnySign4PC or relevant browsing exposure, hunt from the exposure path and browser-to-local-client execution behavior rather than relying only on malware family names; outside that footprint, treat the activity as a regional warning.
Chrome AI vulnerability response priority
Keep the Chrome AI feature vulnerability in the accelerated Chrome patch cycle, especially for high-risk users, but do not spend incident-command time on it unless stronger exploitation evidence appears.
AsyncAPI package supply-chain containment
For projects with affected AsyncAPI tooling in the build or runtime path, freeze npm installs and releases, remove malicious versions, rebuild on clean runners, rotate reachable npm and CI secrets, and verify lockfiles and SBOMs.
Exchange Outlook Web Access containment during reported exploitation
Isolate internet-facing on-premises Outlook Web Access hosts when reported compromise indicators are present or mitigation cannot be verified; otherwise verify mitigation, apply the security update, hunt mailbox and web telemetry, and preserve logs before cleanup.
Build-and-publish trust response for npm and CI exposure
Treat confirmed npm package compromise or build-server exposure as a build-and-publish trust failure. Freeze affected releases, rotate CI, package-registry, cloud, signing, and deployment tokens, audit publish rights and workflow permissions, rebuild important artifacts on clean runners with pinned dependencies, and use SBOMs and lockfiles to identify direct and transitive exposure.
Safe-continuity response for exposed water-sector control paths
For internet-connected PLC, HMI, SCADA, or vendor remote-access paths in water or similar critical-infrastructure environments, treat exposure or suspected intrusion as a safety and continuity incident. Verify process state locally, shift to controlled local or manual operation when needed, isolate external command paths without breaking plant visibility, freeze engineering changes, preserve project files, and rotate engineering and vendor credentials from clean systems.
Joyfill supply-chain CI/CD trust reset
Freeze reported affected Joyfill prerelease consumption and high-privilege deployments; remove malicious packages or workflows where verified; isolate build runners; rotate CI/CD, repository, cloud, registry, and package-publishing secrets from clean systems; and seek vendor provenance assurance before promotion.
FastJson emergency mitigation for exposed services
Verify affected FastJson deployments, prioritize internet-facing services, apply SafeMode or equivalent mitigations, isolate risky endpoints, preserve process and network evidence, hunt for suspicious activity, and rotate secrets reachable by the Java runtime.
Exposed network management patching and compromise hunt
Immediately inventory externally reachable Arista, Fortinet, and VeloCloud management systems; restrict management access; apply available patches or mitigations; preserve and review logs; and hunt for suspicious access. Treat exposed systems or systems with suspicious access as higher risk while exact version and advisory details are verified.
Govern AI agents and SaaS connectors as privileged authority paths
Organizations should govern AI agents and SaaS connectors as privileged authority paths. They should disable or tightly scope agent creation, require admin approval and audit for new agents and connectors, reduce OAuth scopes, use least-privilege service identities, add egress controls, treat retrieved content as untrusted input, strip hidden comments, and require human approval before write, merge, credential, or outbound actions.
Split dependency automation into security and routine lanes
Teams should keep dependency security updates on an expedited path while routing routine version changes through a separate observation lane. They should adjust Python package release workflows for reported file-restriction rules, require lockfiles and hash or digest pinning, and audit publishing tokens and release permissions.
Escalate correlated endpoint blindness as ransomware staging
SOC teams should escalate endpoint heartbeat loss when it coincides with driver loads or security-tool tampering as a high-risk ransomware-staging pattern, collect corroborating telemetry outside the endpoint agent, prioritize vulnerable-driver blocking for high-risk Windows rings, and quarantine suspicious silent endpoints through network or identity controls.
Restrict Microsoft Entra OAuth Device Code Flow
Block Microsoft Entra OAuth Device Code Flow where there is no known business dependency. Where it is needed, limit use to named users, named applications, managed devices, expected locations, Conditional Access, monitoring, and rapid session or refresh-token revocation for suspicious use.
Contain internet-exposed OT access paths
Operators should treat directly internet-exposed PLC, HMI, SCADA, and engineering access paths as untrusted, remove exposure immediately, preserve project evidence, compare controller logic and project files with known-good backups, independently verify HMI and process values, and avoid logic changes without OT owner approval.
Controls for privileged AI and MCP-style agents
Disable or tightly constrain AI review and MCP-style agents when they can act on repositories, pipelines, secrets, wikis, or work items. Strip hidden untrusted content from context, log tool calls, rotate reachable secrets, and issue dedicated least-privilege agent identities instead of using ambient human authority.
Crypto losses separated by control failure
Do not treat the Allbridge, AFX Trade, Lien Finance, and Triple-A losses as one contagion event. Classify each loss by the apparent control failure and act quickly: exchanges freeze or flag known inbound flows, DeFi teams pause only affected pools, bridges, minters, or markets and preserve state, and payment gateways rotate hot-wallet keys and suspend automated sweeps until reconciled.
DevMan and Funky Mantis ransomware readiness posture
Do not change the OT threat model around an unverified industrial locker claim. Monitor affiliate tooling and hunt Windows, Linux, and ESXi ransomware behavior while hardening identity, backup, virtualization, and IT-to-OT recovery choke points.
PLC and HMI manipulation process-integrity response
Reported PLC and HMI manipulation should be handled as a safety and process-integrity response: remove direct internet access for controllers, HMIs, engineering workstations, and vendor remote paths; freeze risky changes; preserve evidence; compare running logic with known-good baselines; and validate display state through independent checks.
PTC Windchill and FlexPLM exposure compromise assessment
Internet-facing PTC Windchill or FlexPLM systems should be removed from direct exposure, validated for applicable vendor-approved fixes or mitigations, preserved for evidence, and assessed as potentially compromised until web-shell artifacts, logs, file integrity, and outbound-transfer paths are reviewed.
Patch and isolate Johnson Controls building-security servers
Facilities using affected building-security application servers should identify exposed or remotely reachable instances, restrict access to trusted admin networks or VPN, and schedule an emergency upgrade to the vendor-confirmed fixed release while deferring broader physical-security redesign.
Restrict crypto bridge operations until controls are proven
Affected crypto bridge protocols should pause or restrict operations until validator keys are rotated and custody is hardened, quorum independence is proven, bridge messages are domain-separated and fully bound, destination payouts are tied to verified source-chain events, and emergency controls are in place.
Limit OAuth apps and AI agent delegated authority
Disable unapproved OAuth apps and broad AI agent connectors, revoke grants and refresh tokens where exposure exists, require fresh explicit authorization with admin policy checks, issue scoped short-lived identities, and require semantic confirmations for high-risk actions.
Showing 30 records · page 4 of 9
These controls filter only the 30 records on this page. Search the full Ledger →