9 verified findings from 1 run · the settled record for this UTC day, in the classic brief order.
Criticality
Kind
Topic
Region
TL;DR · the day in one read
01Mastra npm supply-chain compromise (easy-day-js). Deep dive: the Mastra AI framework's entire npm namespace was backdoored. A trojanised easy-day-js look-alike dependency was swept as a production dependency into 140+ @mastra/* packages in under 90 minutes, delivering a cross-platform credential/wallet stealer; the publishing-account access vector is not disclosed by the primaries (JFrog, 2026-06-17). →
02CVE-2026-0647 et al. Rockwell Automation FLEX I/O unauthenticated password reset (CVSS 9.4) and Logix CIP denial-of-service, flagged by NCSC-CH. Rockwell FLEX I/O adapters: unauthenticated web-interface password reset (CVE-2026-0647, CVSS 9.4), flagged by NCSC-CH. A crafted HTTP GET resets the admin password on 1794-AENTR/AENTRXT EtherNet/IP adapters; companion CVEs crash Logix controllers via malformed CIP (CISA ICS-CERT, 2026-06-16). Fixed in firmware 2.013; segment OT now. →
03CVE-2026-46978 / CVE-2026-35278, Oracle June 2026 CSPU: unauthenticated Solaris RAD flaw (CVSS 10.0) and PeopleSoft RCE (9.8). Oracle June 2026 Critical Security Patch Update ships 245 fixes, ~100 remotely exploitable without authentication. The standouts: CVE-2026-46978 (Solaris 11.4 Remote Administration Daemon, CVSS 10.0) and CVE-2026-35278 (PeopleSoft PeopleTools Performance Monitor, CVSS 9.8), both unauthenticated (SecurityWeek, 2026-06-17 · Oracle, 2026-06-17). No confirmed exploitation yet, patch internet-facing tiers first. →
04ScarCruft (APT37) delivers NarwhalRAT behind fake Microsoft OTP "security alert" lures. ScarCruft (APT37) deploys NarwhalRAT behind fake Microsoft OTP alerts; China arrests 67 Silver Fox/ValleyRAT operators. North Korean spearphishing impersonating Microsoft MFA notices delivers a compiled-Python RAT with a pCloud dead-drop resolver (Genians, 2026-06-16); separately, Chinese police dismantled the supply chain behind the Winos/ValleyRAT operator network. →
05FBI and Secret Service: FortiBleed is still active, locks admins out of FortiGates and feeds ransomware affiliates. FortiBleed is an active credential-compromise campaign against internet-facing FortiGate firewalls and SSL VPN gateways, with SOCRadar counting more than 86,644 compromised devices in 194 countries; Fortinet said in June that it uses no new vulnerability, only reused credentials and brute force. A joint FBI and U.S. Secret Service advisory of 2026-10-06 says attackers keep scanning exposed devices with previously obtained credentials, create administrator accounts on compromised firewalls and in some cases delete or change the original accounts so owners are locked out, and that access brokers using the chain have supplied INC/Lynx and Payload ransomware affiliates. An organisation with an internet-exposed FortiGate should end all administrator and VPN sessions, reset every credential and look for accounts and API keys it did not create. →
Genians Security Center attributed a new campaign to ScarCruft / APT37 (North Korea nexus) deploying a previously-undocumented RAT it calls NarwhalRAT (Genians, 2026-06-16). The lure is a spearphishing email impersonating a Microsoft multi-factor authentication / OTP security alert; the attached ZIP carries a Windows shortcut (LNK) that launches PowerShell with -ExecutionPolicy Bypass to pull a batch loader, which establishes persistence via a scheduled task running on a one-minute interval (T1053.005). The payload is a compiled-Python binary loading obfuscated bytecode and providing keylogging (T1056.001), screenshot and audio capture, USB collection and remote command execution; C2 resilience comes from a pCloud dead-drop resolver (T1102.001) that hands out current relay addresses, defeating static domain/IP blocking (The Hacker News, 2026-06-17).
Why it matters to us: APT37 targets government, diplomatic, policy-research and Korean-diaspora organisations, including in Europe. The behavioural chain is hunt-friendly without IOCs: alert on schtasks.exe creating tasks under an unusual Microsoft…-style name from a non-installer parent, on LNK→PowerShell -ExecutionPolicy Bypass execution trees, and on compiled-Python process images making outbound calls to consumer cloud-storage APIs. Treat the cloud dead-drop pattern as the durable detection surface, blocking one relay does not break C2.
A dataset branded "FortiBleed" was reported on 2026-06-18: researcher Bob Diachenko found a server holding what appeared to be valid Fortinet VPN credentials, Hudson Rock counted 73,932 unique firewall URLs across 194 countries affecting 21,632 unique domains, and Kevin Beaumont, who confirmed the authenticity of some of the admin logins and passwords, put the dataset at roughly 75,000 Fortinet devices (BleepingComputer, 2026-06-18). Fortinet's position is that this is not a new Fortinet vulnerability: it believes threat actors are reusing credentials from previous incidents and using brute force against devices with weak password hygiene and no multi-factor authentication (Fortinet PSIRT, 2026-06-19). BleepingComputer reports researcher Bob Diachenko's claim that a Russian-speaking multi-operator group harvested the credentials, cracked intercepted SSL VPN hashes on a 45-GPU cluster and moved laterally into internal Active Directory environments, with several organisations fully compromised; Diachenko says he drew this from files the attackers left exposed on the same server (BleepingComputer, 2026-06-18); Arctic Wolf is separately tracking the FortiBleed campaign's reach across 194 countries (Arctic Wolf, 2026-06-17). The technique class is valid-account abuse following credential access, not exploitation of a fresh CVE.
which abuses FortiOS's built-in diagnose sniffer packet functionality to capture authentication traffic traversing compromised FortiGate devices
we believe the activity involves threat actors reusing credentials from previous incidents and employing brute-force techniques against devices with weak password hygiene and no multi-factor authentication (MFA)
The FBI says that in some incidents, the threat actor creates administrator accounts and uses their privileges to delete existing admin accounts or change their passwords, denying victims access to their devices.
Some groups benefiting from this are INC/Lynx ransomware and Payload ransomware.
The FortiBleed SSL VPN credential-harvesting campaign has grown from the 73,932 firewall URLs reported on 2026-06-18 (BleepingComputer, 2026-06-18) to 86,644 confirmed compromised credentials across 194 countries, and CISA has published a hardening alert (SecurityWeek, 2026-06-19; CISA, 2026-06-18).
The new detail is methodology and impact, as researcher Bob Diachenko claims it: the operators intercept SSL VPN authentication, crack the hashes on a 45-GPU cluster managed through Hashtopolis and pivot into internal Active Directory environments (SecurityWeek, 2026-06-19; BleepingComputer, 2026-06-18). CISA's alert urges terminating active sessions and resetting credentials, storing administrator logins with PBKDF2, reviewing logs, enabling phishing-resistant MFA and locking down management access (SecurityWeek, 2026-06-19). Fortinet advises checking for unexpected administrator access from an unknown IP and reviewing domain controller logs for lateral movement (Fortinet PSIRT, 2026-06-19).
New analysis published 2026-06-22 gives a more detailed description of the tool chain behind the FortiBleed credential-harvesting campaign. SOCRadar's report, as BleepingComputer describes it, alleges a purpose-built Golang tool, FortigateSniffer, that abuses FortiOS's built-in diagnose sniffer packet command to capture authentication traffic on a compromised FortiGate, with a component named SNIFTRAN reconstructing the captured traffic into PCAP files and a Python toolkit extracting cleartext credentials, password hashes, Kerberos tickets and NTLM authentication material from them, across 24 protocols (BleepingComputer, 2026-06-22).
Fortinet's PSIRT response confirms the campaign uses no new vulnerability: it reuses credentials from the previously-disclosed CVE-2026-24858, CVE-2025-59718 and CVE-2025-59719 plus brute force against devices lacking strong passwords and MFA (Fortinet PSIRT, 2026-06-19; SecurityWeek, 2026-06-22). Reported tradecraft includes a distributed 36-GPU cluster rented from a GenAI company for offline cracking of the harvested hashes, according to researcher Kevin Beaumont (BleepingComputer, 2026-06-22), where Diachenko's earlier claim was a 45-GPU cluster managed through Hashtopolis (BleepingComputer, 2026-06-18); SOCRadar characterises the operators as Russian-speaking (SOCRadar, 2026-06-16).
The delta for defenders is a concrete detection surface that earlier coverage lacked: the tool connects to the FortiGate over SSH and launches the diagnose sniffer packet command (BleepingComputer, 2026-06-22), so hunt for unexpected invocations of that command on the appliance; because harvested Active Directory credentials are the downstream prize, any directory account a FortiBleed-corpus device authenticated against deserves to be treated as exposed, with domain controller logs reviewed for unusual access by those accounts (Fortinet PSIRT, 2026-06-19). Upgrade to firmware with PBKDF2 password hashing to make offline cracking expensive, terminate active sessions, enable MFA and disable external management access.
A joint advisory of the FBI and the U.S. Secret Service (JCSA-20261006-01, 2026-10-06) calls FortiBleed an active, global credential-compromise campaign against internet-facing FortiGate firewalls and SSL VPN gateways, cites SOCRadar for more than 86,644 compromised devices in 194 countries, and says attackers are continuing to scan exposed devices with previously obtained credentials (FBI and U.S. Secret Service, 2026-10-06). The operators scan exposed SSL VPN portals, run credential stuffing and password spraying from earlier Fortinet leak dumps and infostealer logs, pull password hashes and session tokens from compromised devices and crack the hashes offline with Hashcat and Hashtopolis on a rented GPU cluster; the advisory ties the success to reused or leaked credentials and a legacy SHA-256 password storage (FBI and U.S. Secret Service, 2026-10-06).
What the advisory adds is the lockout. On a compromised firewall the actors create administrator accounts that were not there before and, in some cases, delete existing accounts or change their passwords, so the owners cannot log in and recovery needs steps beyond patching and password resets (FBI and U.S. Secret Service, 2026-10-06; BleepingComputer, 2026-10-07). Cracked credentials are enriched and validated by scripts that filter out honeypots and rank targets by revenue and network structure; with verified credentials the attackers enumerate Active Directory accounts and spray passwords to find privileged users, and the operation packages working VPN configurations and target lists for sale, the role of an initial-access broker (FBI and U.S. Secret Service, 2026-10-06). The advisory says the chain has been an initial entry point for ransomware affiliates, currently INC/Lynx and Payload (FBI and U.S. Secret Service, 2026-10-06); SOCRadar had attributed FortiBleed to the Lynx / INC ransomware group in a 29 June update to its report (SOCRadar, 2026-06-16), after reaching both groups' negotiation panels on a server used in the campaign (BleepingComputer, 2026-10-07).
The advisory's measures: restrict external management through trusted hosts, a local-in policy or no internet administration at all; terminate all administrative and VPN sessions and reset all Fortinet VPN and administrator passwords; require phishing-resistant MFA on remote access and administrator accounts; review users, configuration and REST API keys for entries nobody created, removing unknown keys and refreshing the legitimate ones; enforce PBKDF2 for administrator password storage per Fortinet's guidance for FortiOS 7.2.11 and later; and review firewall, VPN, authentication and domain controller logs for lateral movement (FBI and U.S. Secret Service, 2026-10-06). On the FortiGate itself, in analyst judgement, the administrator accounts, passwords and REST API keys the advisory says to review are recorded in the device's configuration and administrator audit log, which is where an unexpected change shows.
The FBI and Secret Service advisory tells operators to review firewall, VPN, authentication and domain controller logs and to check users, configuration and REST API keys for entries nobody created (FBI and U.S. Secret Service, 2026-10-06); it does not name specific Windows event ids or a domain-wide credential rotation, which earlier hunting guidance here attached to the campaign. The mapping of those checks to the FortiGate's own configuration and administrator audit log is a reading of the advisory, not part of it.
The first FortiBleed figures come from three researchers. Bob Diachenko found a server holding what appeared to be valid Fortinet VPN credentials, Hudson Rock counted 73,932 unique firewall URLs across 194 countries affecting 21,632 unique domains, and Kevin Beaumont, who confirmed the authenticity of some of the admin logins and passwords, put the dataset at roughly 75,000 Fortinet devices (BleepingComputer, 2026-06-18). SOCRadar's attribution of the campaign to the Lynx / INC ransomware group dates from a 29 June update to its report (SOCRadar, 2026-06-16). Contradiction: BleepingComputer's October recap dates that link to July (BleepingComputer, 2026-10-07). The FBI and Secret Service advise reviewing firewall and VPN users and other configuration for unauthorized changes, preferably against a well-known configuration, with particular attention to the addition of unrecognized accounts (FBI and U.S. Secret Service, 2026-10-06).
Chinese police arrested 67 suspects across five provinces in a June 2026 operation against Silver Fox (also tracked as Void Arachne, UTG-Q-1000 and TA4922) assessed as one of the most active crimeware operations targeting Chinese-speaking users (Risky Biz News, 2026-06-17). The arrests reportedly span the full criminal supply chain: the primary developer/seller of the Silver Fox (Winos) trojan, a variant developer, phishing-site operators, and fake-app download-site operators, with secondary RATs including ValleyRAT used for credential theft. A CNCERT/CC security alert issued on 2026-05-22 preceded the operation (CNCERT/CC, 2026-05-22).
Rockwell Automation disclosed five ICS CVEs on 2026-06-16, consolidated by NCSC-CH on 2026-06-17 (NCSC-CH Security Hub, 2026-06-17). CVE-2026-0647 (CVSS 9.4) lets an unauthenticated attacker reset the admin password on 1794-AENTR / 1794-AENTRXT FLEX I/O EtherNet/IP adapters (firmware ≤ V2.012) by sending a crafted HTTP GET to the adapter's embedded web server, enabling full takeover and I/O disruption (T0866) (CISA ICS-CERT, 2026-06-16). Companion CVE-2026-0646 (CVSS 7.5) is a CIP-handling DoS on the same adapter requiring a manual reset; CVE-2026-11317 (CVSS 7.5) causes a major non-recoverable fault on CompactLogix/ControlLogix 5370/5570 controllers via a crafted CIP message, requiring a full program download to recover (T0814) (CISA ICS-CERT, 2026-06-16); and CVE-2025-13036 (CVSS 7.7) is an authentication bypass in FactoryTalk Historian Site Edition. FLEX I/O fixes ship in firmware 2.013 (Rockwell SD1775); exploitation status is unknown for all. Where firmware cannot be applied immediately, restrict CIP and HTTP/HTTPS access to these devices to engineering workstations via OT segmentation.
BSI CERT-Bund advisory WID-SEC-2026-1981 (2026-06-17) rates the aggregate severity of the Zammad 7.1 release as "hoch" (high): an attacker can chain the patched flaws to gain administrator privileges, bypass security controls, manipulate or disclose data, or trigger denial-of-service (BSI CERT-Bund, 2026-06-17). Zammad (a widely-deployed open-source helpdesk/ticketing system common in German, Austrian and Swiss public-sector IT service desks) released version 7.1 on 2026-06-16 addressing 13 issues now tracked exclusively as GitHub Security Advisories (Zammad, 2026-06-16); individual CVE identifiers are not yet enumerated in public NVD/CSAF records. Any admin-privilege path in a ticketing system exposes internal IT operations data and staff credentials; internet-exposed instances behind a reverse proxy are highest risk. Upgrade to 7.1 and hunt Zammad audit logs for unexpected role escalations and admin-API calls (e.g. to role/user-management endpoints) from unprivileged sessions.
Oracle's June 2026 Critical Security Patch Update shipped 245 fixes on 2026-06-17, ~100 of them remotely exploitable without authentication (SecurityWeek, 2026-06-17 · Oracle, 2026-06-17). The two standouts for this audience are both pre-auth: CVE-2026-46978 (CVSS 10.0) in the Oracle Solaris 11.4 Remote Administration Daemon (RAD), reachable by an unauthenticated attacker over its default HTTPS management interface, and CVE-2026-35278 (CVSS 9.8), a missing-authentication RCE in PeopleSoft PeopleTools 8.61/8.62 Performance Monitor (T1190). Oracle reports no in-the-wild exploitation at publication; the unauthenticated network vectors warrant emergency prioritisation. Patch internet-facing PeopleSoft and middleware tiers first; as interim hardening, scope the Solaris RAD daemon to localhost where remote administration is not required.
Aikido Security documented a coordinated campaign of at least 15 IDE plugins published under seven vendor accounts on the JetBrains Marketplace between October 2025 and June 2026, posing as AI coding assistants (built on DeepSeek, OpenAI, SiliconFlow) with roughly 70,000 combined installs (Aikido Security, 2026-06-16). The plugins function as advertised but hook the plugin settings-save handler so that the moment a user enters an AI provider API key and clicks Apply, the credential is exfiltrated to an attacker-controlled server; stolen keys are then resold as discounted "paid-tier" access while the legitimate owner pays the bill (Infosecurity Magazine, 2026-06-17). The two largest plugins (CodeGPT AI Assistant, DeepSeek AI Assist) account for most of the ~70,000 installs. Maps to T1195.001 and T1552.001 (credentials in IDE storage). Defenders should not assume the plugins have been removed from the Marketplace, inventory JetBrains plugin installs across developer fleets, rotate any AI provider keys entered into an AI-assistant plugin since October 2025, and move to IDE plugin allowlisting where possible.
Check Point Research detailed a Rust-based clipboard-hijacker campaign against cryptocurrency users whose distinguishing feature is the systematic manipulation of security-tool reputation signals (Check Point Research, 2026-06-17). The operator runs a network of GitHub ghost accounts, SourceForge pages with inflated download counts, AI-narrated YouTube channels and Telegram channels advertising fake crypto "edge" tools (Solana/Pump.fun sniper bots, Aviator predictors), funnelling victims through a WordPress phishing site to download the Rust payloads for Windows and macOS. Critically, the actor submits fake benign community votes and comments on VirusTotal to lower the apparent threat score, so triage analysts relying on community reputation see the sample as pre-vetted. The payload watches the clipboard for wallet-address patterns and silently substitutes attacker addresses. The operational takeaway for SOC triage: VirusTotal community votes/comments are not a trust signal for this malware class, weight first-party engine verdicts and behaviour, and add clipboard-modification (T1115) hooks plus Rust binaries executing from user Downloads/Temp without code-signing to hunt hypotheses.
On 2026-06-17 the entire npm namespace of Mastra (an open-source JavaScript/TypeScript framework for building AI applications, with roughly 1.1 million combined weekly downloads) was backdoored through a single poisoned transitive dependency (JFrog, 2026-06-17 · Socket, 2026-06-17). This is a clean worked example of the failure mode that matters most for any organisation consuming open-source AI tooling: trust in a transitive dependency turns one compromised publishing path into ecosystem-wide code execution on developer and CI machines.
Access vector. The malicious easy-day-js and the wave of @mastra/* republishes were pushed through the project's npm publishing chain; the cited primaries (JFrog, Socket) document the result but do not disclose how the publishing account was obtained, so the brief makes no claim about the initial-access vector (JFrog, 2026-06-17). What matters operationally is downstream regardless of vector: a trusted scope published code that executed on every consumer at install time.
The dependency-substitution chain. Rather than poisoning a Mastra package directly, the attacker moved the malicious behaviour one level down into a new dependency named easy-day-js, a trojanised look-alike of the popular dayjs date library. A clean version was published first so the semver caret range looked benign, then the malicious easy-day-js@1.11.22 was published; an automated wave added it as a production dependency across 140+ @mastra/* packages, with the malicious versions published between roughly 01:15 and 02:36 UTC, under 90 minutes (Socket, 2026-06-17). The two-stage timing is a deliberate attempt to defeat naive dependency-pinning checks. Maps to T1195.002 (Compromise Software Supply Chain) layered on T1195.001 (Compromise Software Dependencies and Development Tools).
Execution and second stage. The malicious package carries a postinstall lifecycle hook (node setup.cjs) that runs automatically during npm install / npm ci (T1059.007, JavaScript). The stage-1 loader disables TLS certificate validation (NODE_TLS_REJECT_UNAUTHORIZED=0), writes marker files to the OS temp directory, downloads a stage-2 Node.js payload, spawns it as a detached hidden process, and deletes setup.cjs to frustrate static analysis (JFrog, 2026-06-17). The stage-2 is a cross-platform (Windows / macOS / Linux) backdoor that beacons host identity and enumerates installed crypto-wallet browser extensions and saved-credential stores, then polls a C2 for follow-on shell/Node commands (T1071.001, Application Layer Protocol: Web).
Persistence, platform-specific, NVM/Node-masquerading. Stage-2 installs persistence tailored to the OS: a per-user LaunchAgent on macOS (T1543.001, Launch Agent), a systemd user service on Linux (T1543.002, Systemd Service), and an HKCU\…\CurrentVersion\Run key on Windows (T1547.001; Registry Run Keys). The labels masquerade as Node Version Manager / Node tooling, a useful hunt concept rather than a hardcoded indicator: persistence entries that look like NVM/Node housekeeping but point at scripts under a user profile or ProgramData path are the tell.
Detection concepts (no IOCs). Hunt for node processes spawned from the OS temp directory (Sysmon EID 1 with parent node/npm/npx and an image path under %TEMP% or /tmp); for new per-user persistence (LaunchAgent / systemd user unit / HKCU Run key) created by a node parent immediately after a package install; and for npm/node processes making outbound TLS where certificate validation has been disabled. Reputable package-security tooling flagged easy-day-js within minutes of publication, so dependency-scanning telemetry is a high-signal early-warning surface.
Hardening. Run npm ls easy-day-js across all workspaces and CI runners and remove the dependency; treat any host that installed an affected @mastra/* version in the exposure window as compromised and rotate all secrets, tokens and wallet material present on it. Structurally: enforce --ignore-scripts (or vetted allowlists) for install-time lifecycle hooks in CI, require lockfile hash/integrity verification and npm provenance attestation, and as general supply-chain hygiene audit npm org membership so publish/maintainer rights stay scoped to active maintainers.
The deep dive on 2026-06-18 documented the easy-day-js poisoning of 140+ @mastra packages but noted the cited primaries did not disclose how the publishing account was obtained, and made no attribution. Microsoft Threat Intelligence has now closed both gaps: it attributes the operation to North Korea's Sapphire Sleet (BlueNoroff / UNC1069) and states the access vector was a dormant former-contributor npm account (ehindero) whose publish rights across the entire @mastra scope were never revoked (BleepingComputer, 2026-06-20).
Microsoft's analysis details the post-install chain, easy-day-js disables TLS verification, pulls a cross-platform Node.js implant that enumerates 166 cryptocurrency-wallet browser extensions and steals browser profiles, then establishes a scdev svchost service running as SYSTEM for boot persistence (Microsoft Threat Intelligence, 2026-06-17). Snyk independently confirms the dormant-account root cause and notes npm does not expire scope-publish permissions on inactivity (Snyk, 2026-06-16). The defender action shifts from "remove easy-day-js" to a structural control: audit your own private-registry and package-scope ACLs for dormant accounts with retained publish rights, and enforce time-bound or MFA-gated publish tokens. Microsoft notes this is Sapphire Sleet's second npm scope-takeover of 2026 (after Axios in April), a systematised dormant-high-privilege-account hunt, not a one-off.
Upgrade Rockwell FLEX I/O adapters to firmware 2.013 and segment OT (§ 2). For CVE-2026-0647 and the Logix CIP DoS CVEs, restrict CIP and HTTP/HTTPS to engineering workstations until firmware is applied.
Audit every internet-facing FortiGate for administrator accounts and REST API keys nobody created, then terminate all administrator and VPN sessions and reset every administrator and VPN password.
Patch the Oracle June 2026 CSPU, internet-facing tiers first (§ 2). Prioritise the unauthenticated Solaris RAD flaw (CVE-2026-46978, CVSS 10.0) and PeopleSoft Performance Monitor (CVE-2026-35278, CVSS 9.8); interim-scope the Solaris RAD daemon to localhost where remote admin is not needed.
2026-06-18-aa7ee817· Anthropic Claude (specific model not determined) · 9 entries published
Items dropped:
A claimed Microsoft Defender Antivirus elevation-of-privilege zero-day ("RoguePlanet", with an alleged CVE id, public PoC and no patch), surfaced by S1, but none of its cited URLs (BleepingComputer, MSRC, SecurityWeek) appear in this run's URL-liveness ledger, and a Phase 2 spot-check of the cited BleepingComputer article returned HTTP 404. Unable to confirm any source was actually fetched; treated as unverified / likely fabricated and dropped per the zero-LLM-knowledge rule. The CVE id is deliberately omitted here pending an independently verifiable advisory. If a genuine Defender EoP zero-day with public PoC is confirmed, it returns next run.
DragonForce "Backdoor.Turn" (Microsoft Teams TURN-relay C2), surfaced by S4 but already the 2026-06-17 deep dive; no in-window material delta beyond that coverage. Dropped (BYOVD/Teams-relay hardening retained as an action item in the prior brief).
Sophos CTU "AI in the underground", single-source trend/awareness item with no specific technique, CVE or detection hook; dropped under less-is-more.
Correction applied during verification (FortiBleed, § 1): S1's research draft over-stated the framing, describing FortiBleed as "73,932 FortiGate admin credential sets" leaked via an old FortiOS authentication-bypass vulnerability chain, and citing a fabricated Fortinet PSIRT URL (FG-IR-26-FortiBleed). Corrected against the primaries: it is a credential exposure of 73,932 device URLs (~75,000 devices, 194 countries) assembled from brute-force and reshared prior-incident data, not a new vulnerability (Fortinet's own statement). The fabricated PSIRT URL was removed and the item re-anchored to the two ledger-verified sources. Sourcing precision: the Russian-speaking-actor / Active-Directory-lateral-movement detail is supported by BleepingComputer; Arctic Wolf supports the 194-country campaign reach (Arctic Wolf separately describes a SHA-256→PBKDF2 password-hash-storage weakness and an associated FortiOS CVE, which this brief does not rely on).
Zammad (§ 2): individual CVE identifiers for the 13 June 2026 GitHub Security Advisories are not yet enumerated in public NVD/CSAF; the item is sourced to the BSI advisory and the Zammad release and carries no CVE pill by design.
Reduced confidence:China, Silver Fox arrests (§ 1) is MEDIUM confidence, the primary (Risky Biz News) summarises Chinese-language law-enforcement reporting, corroborated by the CNCERT/CC advisory; EU nexus is indirect (diaspora-targeting lures).
Single-source items: none beyond the national-CERT / primary-research carve-out.
Deliberate non-inclusion (Oracle / ShinyHunters): verification noted that SecurityWeek's June 2026 CSPU coverage also references the separately-tracked ShinyHunters exploitation of Oracle PeopleSoft/E-Business Suite (CVE-2026-35273) against many organisations. That campaign is an already-covered ongoing story (multiple prior briefs and the 2026-W24 weekly); no verified fresh in-window delta surfaced this run, so it is not re-reported here. The § 2 Oracle item intentionally covers the new June CSPU criticals (CVE-2026-46978, CVE-2026-35278), which are not yet exploited.
Contradictions: none material this run.
Source list: added aikido-security as a candidate (software supply-chain / IDE-security research; primary for the JetBrains plugin disclosure, § 3). One-candidate cap respected.
Sub-agents: all four (S1–S4) returned within budget; all reported Claude Sonnet 4.6.
Coverage gaps: inside-it-ch (Cloudflare challenge; no usable Wayback snapshot, Swiss regional IT news missed); enisa-news-rss (HTTP 404; ENISA EUVD bridge used, no in-window criticals); cert-fr-actu (feed stale since Nov 2025); databreaches-net (HTTP 403, no Wayback snapshot, covered via alternates); sophos-xops (fetched OK, one item used); oracle-cpu (HTTP 403, covered via SecurityWeek/Oracle CSPU/NCSC-NL); projectzero, greynoise, elastic-seclabs, dfirreport, msft-secblog, compass-security, sec-disclosures-edgar, edpb, ico-uk, no in-window qualifying items.
Unmatched action items (migrated)
Treat every internet-exposed FortiGate's admin/VPN credentials as exposed and rotate now (§ 0, § 1 FortiBleed). Force admin and VPN password resets, enforce MFA on all administrative/VPN logins, take the management interface off the WAN, and review FortiGate admin-login events plus domain-controller authentication (Windows EID 4624/4768) for logins from unexpected source addresses. Patching does not rotate a leaked credential.
Upgrade Zammad to 7.1 and hunt for admin-role escalation (§ 2). Review Zammad audit logs for unexpected role changes and admin-API calls from unprivileged sessions; gate internet-exposed instances behind VPN/mTLS.
Run npm ls easy-day-js across all workspaces and CI runners; treat affected hosts as compromised (§ 5). Remove the dependency, rotate secrets/tokens/wallet material on any host that installed an affected @mastra/* version, enforce --ignore-scripts + lockfile integrity in CI, and automate publish-access revocation on contributor offboarding.
Inventory JetBrains plugins and rotate AI provider API keys entered into any AI-assistant plugin since October 2025 (§ 3); move toward IDE plugin allowlisting.
Stop treating VirusTotal community votes/comments as a trust signal in SOC triage for fake-tool malware (§ 3); weight first-party engine verdicts and behaviour.