Executive Summary
In November 2026, threat actors compromised Ethersphere's CI/CD pipeline using stolen automation credentials to publish two malicious releases: bee-js 13.3.0 (live ~75 minutes on npm) and Swarm Desktop 0.56.2 (live ~8.5 hours on GitHub). Both packages delivered a loader that exfiltrated Swarm node wallet keys and plaintext passwords to attacker-controlled infrastructure, then installed persistent backdoors—a Go remote shell (evjawbreaker) on Linux and macOS and a ScreenConnect remote-access client on Windows. The backdoor's command-and-control address was read from an Ethereum smart contract, enabling dynamic server rotation. Linked to the earlier SleepyDuck and EtherDuck campaigns, the attack targeted the Web3/Swarm developer ecosystem with potential cryptocurrency theft.
This incident illustrates the accelerating weaponization of open-source package registries and CI/CD pipelines against blockchain and Web3 developers. As supply chain attacks grow more sophisticated—leveraging valid code-signing certificates, provenance attestations, and auto-update mechanisms—organizations must enforce egress controls, runtime behavioral monitoring, and strict approval gates on release workflows to contain blast radius.
Why This Matters Now
Open-source supply chain attacks are surging in 2025–2026, with threat actors specifically targeting Web3 and crypto developer toolchains. Valid code signatures and provenance records no longer guarantee safety. Regulators and frameworks (NIST SSDF, EO 14028) now mandate software supply chain controls, making this incident a direct compliance and operational risk signal for any organization consuming npm or GitHub-distributed packages.
Attack Path Analysis
Attackers stole CI credentials and secrets from the Ethersphere project, gaining access to the bee-worker GitHub automation account and a maintainer account to inject malicious code into bee-js 13.3.0 and Swarm Desktop 0.56.2 via the project's own trusted release pipelines. The malicious releases downloaded a second-stage payload from a threat-actor-controlled domain, installed persistent backdoors across all major OS platforms, exfiltrated Swarm node wallet keys and plaintext passwords, and established durable remote access enabling ongoing theft of cryptocurrency assets.
Kill Chain Progression
This analysis maps confirmed threat intelligence to the full cloud kill chain to show where defensive gaps would emerge as an attack progresses.
Initial Compromise
Description
Attackers stole CI credentials and secrets — including GitHub tokens for the bee-worker automation account and a long-time maintainer account — granting them authenticated write access to the Ethersphere release pipeline. Using these valid credentials, they triggered GitHub Actions workflows without pull request review or approval gates, injecting malicious postinstall scripts into bee-js 13.3.0 and a malicious loader into Swarm Desktop 0.56.2. The packages were signed with the project's legitimate Apple Developer ID and carried valid npm provenance, bypassing trust controls.
MITRE ATT&CK® Techniques
Compromise Software Supply Chain
Valid Accounts: Cloud Accounts (Stolen CI/CD Credentials)
Command and Scripting Interpreter: JavaScript
Create or Modify System Process: Launch Agent
Create or Modify System Process: Systemd Service
Unsecured Credentials: Credentials In Files
Exfiltration Over C2 Channel
Remote Access Software
Potential Compliance Exposure
Mapping incident impact across multiple compliance frameworks.
NIST SP 800-53 Rev 5 – Provenance of Software Components
Control ID: SR-4
CISA Zero Trust Maturity Model 2.0 – CI/CD Pipeline Access Controls and Least Privilege
Control ID: Applications Pillar – Automation & Orchestration
NIS2 Directive – Security of the Software Supply Chain
Control ID: Article 21(2)(d) – Supply Chain Security
DORA (EU 2022/2554) – Monitoring and Integrity of ICT Tool Providers
Control ID: Article 30 – ICT Third-Party Risk Management
PCI DSS 4.0 – Software Component Integrity and Patch Management
Control ID: Requirement 6.3.3 – Security of Software Components
ISO/IEC 27001:2022 – Security in Software Development and Supply Chain
Control ID: Annex A 8.30 – Outsourced Development
Sector Implications
Industry-specific impact of the vulnerabilities, including operational, regulatory, and cloud security risks.
Financial Services
Supply chain attack targeting Swarm node wallets exposes xDAI/xBZZ token theft via stolen keystore files, with exfiltration-weighted validated_info confirming critical credential and fund loss risk.
Computer Software/Engineering
Malicious bee-js 13.3.0 postinstall script and compromised CI credentials weaponized npm publishing pipelines, threatening developer environments with backdoor installation across Windows, macOS, and Linux.
Internet
Decentralized Swarm storage infrastructure compromised via stolen release automation credentials, enabling wallet-stealing malware and persistent ScreenConnect backdoors across distributed Web3 node operators.
Information Technology/IT
Compromised CI/CD pipelines with valid Apple notarization and npm provenance bypassed supply chain controls, mapping directly to NIST.800-53.SC-7 and ZTMM lateral movement and exfiltration risk vectors.
Sources
- Hackers Compromise Ethersphere's bee-js and Swarm Desktop to Deliver Wallet-Stealing Malwarehttps://thehackernews.com/2026/11/hackers-compromise-etherspheres-bee-js.htmlVerified
- bee-js Compromise: Supply Chain Attack on Ethersphere npm Packagehttps://socket.dev/blog/bee-js-compromiseVerified
- Ethersphere bee-js npm Compromise - SafeDep Technical Analysishttps://safedep.io/blog/ethersphere-bee-js-npm-compromiseVerified
- GitHub Security Advisory GHSA-hh7j-5m3j-23pf: @ethersphere/bee-js 13.3.0 malicious releasehttps://github.com/advisories/GHSA-hh7j-5m3j-23pfVerified
- Swarm Desktop v0.56.3 Clean Release and Incident Noteshttps://github.com/ethersphere/swarm-desktop/releases/tag/v0.56.3Verified
- Swarm Desktop Incident Response Pull Request - Maintainer Noteshttps://github.com/ethersphere/swarm-desktop/pull/582Verified
- EtherDuck Analysis: Malicious Solidity VSX Extensions on Open VSX Marketplacehttps://pluto.security/blog/sleepyduck-malware-open-vsx-extensions/Verified
Frequently Asked Questions
Cloud Native Security Fabric Mitigations and ControlsCNSF
Based on the attack progression modeled above, these are the defensive controls that would constrain each stage.
Aviatrix Zero Trust CNSF is directly relevant to this supply chain compromise because the attackers relied on unrestricted outbound connectivity, uncontrolled east-west paths within CI environments, and open egress channels to exfiltrate stolen wallet credentials. By enforcing identity-aware segmentation and controlled egress across cloud workloads, CNSF would likely have reduced the attacker's reachability and constrained the blast radius across affected developer and CI environments.
Control: Cloud Native Security Fabric (CNSF)
Mitigation: CNSF fabric-level visibility across CI pipeline workloads could likely surface unexpected outbound call patterns originating from the compromised automation account, reducing the window in which malicious workflow execution goes undetected and constraining how far tainted build artifacts propagate into connected environments.
Control: Zero Trust Segmentation
Mitigation: Zero Trust segmentation policies would likely constrain what a newly escalated process on a compromised workstation could reach within the broader network, reducing the operational value of elevated local privileges by limiting reachability to sensitive internal services or adjacent cloud workloads.
Control: East-West Traffic Security
Mitigation: East-west traffic enforcement across CI and build workload segments would likely constrain how far a compromised runner or developer machine could reach into adjacent CI jobs, internal registries, or downstream build environments, reducing the spread of tainted artifact execution across the pipeline fabric.
Control: Multicloud Visibility & Control
Mitigation: Multicloud visibility and centralized traffic control across workload egress paths could likely surface anomalous outbound connections to unknown external domains and non-standard IP destinations, reducing the reliability of C2 channels that depend on unmonitored egress to operate persistently across cloud-hosted or cloud-adjacent environments.
Control: Egress Security & Policy Enforcement
Mitigation: Egress policy enforcement on cloud workload outbound paths would likely constrain unauthorized HTTP POST traffic destined for unknown external domains, reducing the attacker's ability to reliably transmit stolen wallet credentials and keystore files outside the environment through uncontrolled egress channels.
Where prior CNSF controls constrained C2 channel establishment and restricted egress paths, the operational reach of persistent backdoors would likely be reduced — limiting the attacker's ability to issue ongoing remote shell commands or stage further cryptocurrency theft from affected Swarm nodes, even where the implants themselves remained installed.
Impact at a Glance
Affected Business Functions
- Decentralized Storage Node Operations (Swarm Network Participation)
- Cryptocurrency Wallet and Key Management (xDAI, xBZZ tokens, chequebook, stake)
- npm Package Ecosystem Integrity (bee-js library consumers, swarm-cli, bee-dashboard)
- CI/CD Pipeline and Release Automation Security
Estimated downtime: 1 days
Estimated loss: N/A
Swarm Desktop node private keys (encrypted Ethereum keystore swarm.key) and associated plain-text passwords (config.yaml) were exfiltrated for affected users who ran version 0.56.2 or installed bee-js 13.3.0 with install scripts enabled. Stolen credentials enable full takeover of node wallets including xDAI and xBZZ token balances, chequebook funds, and staked assets. Additionally, a persistent backdoor (evjawbreaker Go remote shell on Linux and macOS; ScreenConnect RAT on Windows) was installed on compromised machines, providing attackers ongoing access to all data and credentials accessible by the compromised user account. Approximately 65 downloads of Swarm Desktop 0.56.2 and roughly 10 downloads of bee-js 13.3.0 were recorded before removal, though exact infection count is unconfirmed. No theft of funds was reported at time of publication.
Recommended Actions
Key Takeaways & Next Steps
- • Enforce Egress Security & Policy Enforcement on CI/CD and developer endpoints: Deploy FQDN-based egress filtering and outbound traffic policy enforcement to block unauthorized connections to attacker-controlled domains such as api.frameworkmetrics[.]xyz. Postinstall scripts and application loaders that attempt to reach unknown external endpoints at install or launch time should be detected and blocked before second-stage payloads are delivered or wallet files are exfiltrated.
- • Apply Multicloud Visibility & Control to detect anomalous outbound patterns: Instrument CI/CD pipeline network flows and developer workstation egress with centralized traffic observability. Anomalous POST requests carrying custom headers (e.g., X-Ethersphere: 1) to newly registered or low-reputation domains, or connections to blockchain RPC endpoints from build environments, should trigger automated alerting and incident response workflows.
- • Leverage Cloud Firewall (ACF) and Inline IPS (Suricata) to block known-bad payloads and C2 channels: Deploy cloud-native egress firewall controls with URL and FQDN filtering to prevent implants from contacting C2 infrastructure. Suricata-based inline IPS with signatures covering the identified IOCs (domain, IP:port pairs, HTTP paths, SHA-256 hashes) would detect and block evjawbreaker callback traffic and ScreenConnect relay connections at the network layer.
- • Implement Zero Trust Segmentation to limit blast radius from compromised CI runners: Apply identity-based microsegmentation policies so that CI runner workloads operate under least-privilege network access, with no lateral reach to developer workstations, internal registries, or production environments. A compromised npm install in one runner should not be able to propagate malicious artifacts or credentials to adjacent systems.
- • Use Encrypted Traffic controls and East-West Traffic Security to protect secrets in transit and detect internal exfiltration: Ensure that credential and key material traversing internal networks (e.g., between CI runners and secret stores) is protected by MACsec or IPsec. Monitor east-west traffic between workloads for unexpected data flows such as config file reads followed by outbound connections, which are indicative of the wallet key and password exfiltration pattern observed in this attack.



