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
Emergency remediation for exposed edge and trust infrastructure
Prioritize patching or isolation for internet-facing enterprise edge and trust infrastructure named in the briefing where exploitation or emergency-remediation pressure is cited; sequence by external exposure and trust role, preserve evidence, and hunt for persistence.
SonicWall SMA1000 emergency remediation
For exposed affected SonicWall SMA1000 appliances, approve controlled downtime if needed, preserve logs and configuration first, move to the fixed releases reported in the packet, and rebuild or redeploy with credential and token resets if compromise indicators appear.
Immediate validation of cloud and SaaS delegated-trust boundaries
Audit Entra logs for suspicious app or client identifiers and token-theft signs, revoke risky sessions and refresh tokens, force reauthentication, audit OAuth grants, check IoT certificate-to-device policy boundaries, review Splunk secret exposure, and remove broad AI-agent production access until reviewed.
Credential-exposure response for malicious developer package installs
Validate developer workstations and CI runners for affected installs. If install-time execution and secret reachability are confirmed, rotate npm, GitHub, cloud, CI, signing, publishing, and wallet credentials; strengthen package-publishing controls and avoid blind enterprise-wide rotation.
Emergency compromise validation for exposed edge and enterprise systems
Treat exposed affected SonicWall SMA1000, on-premises SharePoint, Oracle E-Business Suite, legacy Cisco IOS, and relevant domain-trust paths as requiring potential-compromise validation. Restrict public access or allow-list exposure, preserve logs and configurations, then patch or rebuild under emergency change control; rotate sessions and trust material where investigation supports it and hunt for persistence.
Treat DeFi automation dependencies as custody incidents
DeFi protocols should treat oracle, keeper, vault automation, and bridge dependency failures as custody incidents: pause affected markets or vaults before code rotation, disable keeper and forwarder paths, rotate or revoke oracle signer keys, reject stale or future-dated price reports, validate independently before payouts resume, and coordinate quickly where funds remain traceable.
Default-deny AI coding-agent authority
Do not allow AI coding agents or MCP and server extensions to inherit developer authority by default; require workspace allowlists, log all tool and command calls, update affected tools per vendor guidance, and apply source-code DLP, secrets scanning, and upload restrictions to AI builders.
Reset durable trust after credential and token exposure
A password reset is not enough when durable trust may be exposed; revoke or constrain refresh tokens, active sessions, OAuth consents, device-code flow, MFA recovery objects, SaaS sessions, developer tokens, endpoint keychain secrets, and downstream credentials where exposure is plausible.
Keep LegacyHive below emergency perimeter containment
Do not promote LegacyHive ahead of active perimeter containment; harden and monitor high-value Windows systems, and escalate only if in-the-wild exploitation, endpoint compromise, or weaponized chaining appears.
Contain exposed perimeter and application systems before routine patching
Exposed SonicWall SMA1000 and Ivanti Sentry systems should be restricted or isolated first, with logs and configurations preserved, rapid vendor-guided remediation, and appliance-held secrets revoked; Cisco IOS and ColdFusion systems should be hunted for exposure and isolated if compromise indicators appear.
Triage OT edge and ICS advisories without blind patching
Pull engineering into triage for exposed router and edge paths and packet-flagged ICS/OT advisories. First isolate exposure, restrict management access, confirm backups and configuration exports, add segmentation and monitoring, then patch during approved outages unless active exploitation or unsafe exposure is confirmed.
Prioritize exposed enterprise systems over bulk patching
Prioritize exposed SharePoint Server, SonicWall SMA1000, and vulnerable Joomla deployments for patching or isolation within 24 hours; preserve logs and hunt for web shells, authentication-bypass activity, appliance abuse, and post-exploitation before declaring systems clean. Stage the broader Microsoft patch wave by exposure and business criticality.
Prioritize AI platform risks by exposure to secrets, execution, and automation
Inventory and block public access to MCP, local LLM, JSON-RPC, assistant configuration, and .env paths; validate vendor applicability before patching or isolating affected AI platform deployments; audit Langflow and similar tools for AWS metadata, secrets, CI/CD, and internal API access; rotate keys only if suspicious access appears.
Revoke delegated trust before relying on password resets
Revoke active sessions, refresh tokens, OAuth grants, device-code sessions, suspicious enterprise applications, connected-app access, and exposed cloud keys before relying on password resets; then re-consent under tighter controls and reissue scoped credentials.
Isolate applicable self-managed ShareFile Storage Zone Controllers
Isolate or power down applicable self-managed/on-prem ShareFile Storage Zone Controllers, preserve logs and configuration, and restore only after applicable vendor mitigation is validated or a clean rebuild is completed.
Harden and hunt exposed AI developer endpoints
Inventory MCP, Ollama, and OpenAI-compatible endpoints; bind them to localhost, VPN, or private networks; require auth or mTLS; remove secrets from assistant config files; block agent/tool runtimes from reaching cloud metadata services unless explicitly needed; and hunt external probes and metadata calls from agent hosts.
Split CMS webshell incidents from exposure-only cases
Use an evidence-based split: isolate confirmed compromised CMS hosts with webshell evidence, preserve evidence, rotate related credentials, and rebuild from clean media; if there is only matching exposure without shell evidence, patch and hunt, remove public admin paths, and add WAF/CDN controls.
Revoke delegated identity trust before credential rotation
Treat OAuth grants, sessions, refresh tokens, connected-app tokens, exposed cloud principals, STS sessions, service-account delegation, and guest access as attacker-held trust state; revoke or freeze them before relying on password or key rotation, then re-consent only approved apps and hunt SaaS-to-SaaS access.
Emergency-triage exposed PAN-OS portal deployments
Review exposed PAN-OS GlobalProtect and Captive/User-ID Portal deployments today; restrict or disable exposed portals where appropriate, patch or mitigate per confirmed vendor guidance, and hunt for unauthorized VPN sessions, firewall changes, credential theft, lateral movement, suspicious management logins, and configuration changes.
Isolate affected Progress ShareFile Storage Zone Controllers
Treat affected on-prem Progress ShareFile Storage Zone Controllers as an immediate containment decision where Progress guidance applies: restrict exposure, isolate or power off affected controllers if safe containment is not available, preserve logs and forensic data, and rebuild or restore only after validated guidance.
Treat suspected AiTM and SaaS abuse as session and OAuth trust compromise
For suspected Microsoft 365 AiTM, Salesforce, OAuth, or help-desk social-engineering exposure, handle the event as potential attacker-held session, refresh-token, or connected-app trust rather than password-only compromise. Revoke sessions, invalidate refresh tokens, review OAuth and connected apps, inspect risky sign-ins, Salesforce exports and API volume, review new MFA enrollments or reset events, verify high-risk help-desk resets through pre-registered channels, and enforce phishing-resistant authentication for admin and high-risk roles.
Scope secret rotation and artifact rebuilds to executed supply-chain code
Rotate or revoke secrets and rebuild artifacts only where malicious code executed in a trusted developer, CI, build, release, or privileged browser context with access to secrets or artifacts. Elsewhere, remove and block affected artifacts, pin or block bad versions, disable untrusted lifecycle scripts where feasible, block ModHeader until a clean build is validated, and avoid a blanket engineering freeze for unaffected environments.
Hard-isolate exposed D-Link DIR-513 v1.10 devices
Promptly retire or hard-isolate D-Link DIR-513 v1.10 devices that are exposed or show exploit attempts, remove WAN exposure, restrict management to allowlisted networks, block external access to /goform/formDeviceReboot, enable applicable IPS protection, and verify replacement or firmware options through authoritative sources.
Use scoped containment for developer supply-chain exposure
Scope exposure to affected jscrambler, ModHeader, and poisoned Go-module paths; rotate reachable developer, CI/CD, cloud, signing, registry, browser, and AI-tool secrets where malicious code executed; quarantine affected build paths; pin known-good versions identified by local inventory; rebuild artifacts from clean runners; and add provenance checks.
Force managed Android Chrome updates without full-device RCE claims
Force managed Chrome for Android 150.0.7871.47 or later where applicable, assess managed WebView exposure, enforce EMM compliance and conditional access for high-risk users, and brief the issue as urgent browser/WebView exposure rather than confirmed Android RCE, sandbox escape, or full-device takeover.
Patch Rust libp2p gossipsub separately from Bonzo
For operators that validate affected Rust libp2p gossipsub exposure, prioritize prompt patching for CVE-2026-34219 and keep that CVE separate from the Bonzo Lend oracle-exploit response.
Handle MFA phishing as session and token compromise
For Evilginx-style AiTM or Microsoft device-code phishing exposure, start with trust-state destruction: revoke sessions and refresh tokens, review OAuth grants and enterprise app consent, restrict or disable device-code flow where feasible, hunt for token replay, and prioritize phishing-resistant authentication for sensitive users.
Treat Canonical/Ubuntu hacktivist claims as availability triage absent supply-chain evidence
Treat the Canonical/Ubuntu pro-Iran hacktivist claim as geopolitical disruption context and availability-risk triage, not as assumed Iranian state tasking or Ubuntu supply-chain compromise, unless authoritative evidence shows deeper impact.
Escalate GitHub ghost-account API reconnaissance to source-control response
Enable and stream GitHub api.request audit events, hunt dormant-account enumeration and unusual user agents, correlate enumeration with private-repository cloning or access anomalies, revoke abnormal PAT/OAuth tokens, expire stale PATs, review OAuth grants, and remove dormant external collaborators.
Lock down Entra passkey and device-code abuse paths
Restrict passkey/FIDO2 registration by group and trusted context, block or tightly scope device-code flow, alert on authentication-method changes, remove suspicious enrolled methods, and revoke sessions or refresh tokens where abuse is suspected.
Showing 30 records · page 6 of 9
These controls filter only the 30 records on this page. Search the full Ledger →