New:Introducing Socket Scanning for VS Code Marketplace Extensions.Learn more →
Get Started

Secure OSS Dependencies

Socket for Open Source Security

Quickly evaluate the security and health of any open source package. Unlike a traditional vulnerability scanner, Socket can detect an active supply chain attack and help you block it. Socket detects over 70 issues in open source code for comprehensive protection.

Explore Integrations
Book a Demo

Call us at (844) SOCKET-0

Beyond CVE Scanning

What is Open Source Security?

Open source code makes up more than 90% of modern software projects, with many apps spamming 10,000+ dependencies. This makes it easy for attackers to use open source as a vector for attacks where open source packages registries are frequently the target of malware. Traditional vulnerability scanners cannot detect active supply chain attacks. Socket's free GitHub app safeguards your open source code from both vulnerable and malicious dependencies.

Install GitHub AppContact Sales
Socket Website

We protect you from vulnerable and malicious packages

github.com/nccgroup/singularity

v0.0.0-20261120062225-3f7e94358921

Live on go

Blocked by Socket

The code implements a flood/DoS-like pattern by launching a large number of outbound HTTP requests to dynamically constructed hostnames. It uses a worker-based message handler to receive parameters (hostname, port, iterations) and performs a burst of fetch operations without handling responses. While not exhibiting explicit data exfiltration or backdoors, this behavior is malicious in contexts where it is used to overwhelm a target or violate terms of service.

github.com/jm33-m0/emp3r0r/core

v0.0.0-20250123065341-91279bff2618

Live on go

Blocked by Socket

This code coordinates proxy discovery and sharing among agents, starts SSH/KCP reverse tunnels, and can expose a SOCKS5 proxy on all interfaces. It encrypts broadcast proxy messages, but logs decrypted messages and credential-bearing proxy URLs, which can disclose the configured proxy password. These capabilities are high risk in an unauthorized deployment and are characteristic of command-and-control or remote-access tooling; the fragment alone does not establish operator intent. No external domains, fixed IPs, or file paths are present.

hostinger-api-mcp

4.15.0

by devs_hostinger

Live on npm

Blocked by Socket

The supplied JavaScript is a readable static catalog, not an executable malware payload, and no credential theft or system damage is directly demonstrated. However, its headless-build instructions require sending build reports containing domain, environment, URLs, timing, issues, and user-feedback metadata to hosting_headless_submit-build-report without consent and explicitly direct that this reporting be kept undisclosed until the final report. Relevant indicators include Hostinger reporting via that operation, HOSTINGER_API_TOKEN and UNSPLASH_ACCESS_KEY references, and local paths under ~/.hostinger/<domain>/ and ~/.config/hostinger-mcp/credentials.json; no secret values are included. Severity is elevated for privacy and transparency risk; incomplete script bodies prevent a full runtime assessment.

github.com/nccgroup/singularity

v0.0.0-20261120063628-00a04fbaf541

Live on go

Blocked by Socket

High severity: this is an explicit browser-based DNS-rebinding and local-network access bypass tool, not ordinary application fetch logic. It probes and navigates to a configured target, can execute WebSocket-directed fetches, and relays target response data and accessible cookies to its parent/callback or a WebSocket peer. The WebSocket endpoint is dynamically hostname-derived at ws://<server>:<configured-port>/soows (default port 3129); no fixed external domain or IP is present. The code is not substantially obfuscated, but should only be used in an authorized security-testing environment.

motionframe-plugins

0.0.1-security

by npm

Live on npm

Blocked by Socket

Malicious code in motionframe-plugins (npm) Source: ghsa-malware (74587b11410300a69bb121b82387e8ffe1ed18cc552c818f1ada8fdbc45a7a21) Any computer that has this package installed or running should be considered fully compromised. All secrets and keys stored on that computer should be rotated immediately from a different computer. The package should be removed, but as full control of the computer may have been given to an outside entity, there is no guarantee that removing the package will remove all malicious software resulting from installing it.

layer@layer-app.com

3.5.0

by anonymous-3e1c80576c581932890177638d1e4eb6

Live on firefox

Blocked by Socket

This module implements clipboard event monitoring (copy/paste), extracts selected/pasted text, computes SHA-256 hashes (with a sessionSalt) of the clipboard text and page URL, augments events with detailed DOM field/page context, batches events, and transmits them to a hardcoded remote Supabase endpoint when locally configured (layer_code/host allowlist) and when extension session keys are available. The main concern is privacy-sensitive data collection and outbound exfiltration, even though plaintext clipboard data is not transmitted.

github.com/developing-today/code/pkgs/identity

v0.0.0-20261120163205-96570b03c090

Live on go

Blocked by Socket

The script contains a high-risk remote code execution pattern: it retrieves an init script from a network-controlled charm endpoint, writes it to a local file, and executes it without cryptographic verification. It also transmits “keys” derived from the local charm identity material to the remote endpoint, creating a potential secret exfiltration channel. Additionally, it performs multiple supply-chain affecting operations (apt installs/upgrades, curl+dpkg installer, npm@latest install, and git clone from the internet). While the code is not obfuscated, the trust and execution flow makes it plausibly dangerous and should be treated as high risk in a supply-chain context.

github.com/jm33-m0/emp3r0r/core

v0.0.0-20250123094136-63cd9914d9d0

Live on go

Blocked by Socket

This code coordinates proxy discovery and sharing among agents, starts SSH/KCP reverse tunnels, and can expose a SOCKS5 proxy on all interfaces. It encrypts broadcast proxy messages, but logs decrypted messages and credential-bearing proxy URLs, which can disclose the configured proxy password. These capabilities are high risk in an unauthorized deployment and are characteristic of command-and-control or remote-access tooling; the fragment alone does not establish operator intent. No external domains, fixed IPs, or file paths are present.

hydromind

5.24.8

Live on pypi

Blocked by Socket

This file is a high-suspicion import-time dynamic execution stub: it delegates execution of hydromind.control.query to a native extension helper and runs it inside the current module’s globals(), enabling any in-process behavior implemented in the loaded module or the native helper. This snippet alone does not prove malicious activity (no explicit network/file/credential logic is present here), but the load-time execution pattern is a meaningful supply-chain risk and warrants inspecting hydromind.control.query and hydromind._protection_native for concrete malicious indicators.

hydromind

5.24.8

Live on pypi

Blocked by Socket

This module is an import-time dynamic loader that immediately delegates execution to a native helper to load and execute `hydromind.control.forecast_fusion` while granting it the caller’s `globals()` context. The fragment itself shows no explicit malicious actions (e.g., network/process/credential theft), but the strong indirection through a native extension and the import-time execution pattern are significant supply-chain/security red flags. To assess risk, the native helper (`hydromind._protection_native`) and the loaded target module (`hydromind.control.forecast_fusion`) must be reviewed for actual payload behavior.

github.com/jm33-m0/emp3r0r/core

v0.0.0-20250123094136-63cd9914d9d0

Live on go

Blocked by Socket

This Linux server component provides emp3r0r command-and-control communication, agent check-in, file transfer, reverse-shell streams, and port forwarding, making it high severity if deployed without authorization. It listens on 0.0.0.0 at the configured Shadowsocks port and on the configured C2 TLS port; TLS files are read from EmpWorkSpace/emp3r0r-cert.pem and EmpWorkSpace/emp3r0r-key.pem, and downloads use paths derived through omitted path-generation code under Temp. No external domain, fixed IP, hardcoded credential, or environment-variable harvesting is present in the fragment. The C2 functionality is clear, but this code alone does not prove that its use is unauthorized.

github.com/jm33-m0/emp3r0r/core

v0.0.0-20261120044915-f51c6aa83ed4

Live on go

Blocked by Socket

The fragment implements file I/O, but explicitly designs that I/O to evade Linux syscall-entry monitoring and discusses avoiding an environment-based detection vector. This is a significant defense-evasion indicator and warrants treating the code as high risk; no data exfiltration or other payload is visible in the supplied fragment.

github.com/jm33-m0/emp3r0r/core

v0.0.0-20261120044915-f51c6aa83ed4

Live on go

Blocked by Socket

This is a purpose-built C2 transport component with certificate pinning-to-CA-style trust and optional camouflage/routing features. It does not itself show host-data theft or destructive behavior, but its C2 communication functionality is a strong malicious-use indicator in an otherwise untrusted package. Plain HTTP configuration would expose traffic to interception.

@parcha/agentrun-harness-acp

0.1.0-beta.48

by GitHub Actions

Live on npm

Blocked by Socket

This is a configurable fake ACP/MCP agent and test harness, but its enabled config.leak option can intentionally copy CURSOR_API_KEY (raw or base64-encoded) into a file selected by config.leak.path; config.authFile and config.countFile also permit configurable writes. Path traversal is possible through configured relative paths, while network requests and subprocess execution are driven by configured MCP servers. No fixed external domain or IP is present. Severity is conditional and high if untrusted configuration enables credential leakage; otherwise the shown behavior is primarily harness functionality.

@zaiinlin/bails

0.0.1-security

by npm

Live on npm

Blocked by Socket

Malicious code in @zaiinlin/bails (npm) Source: ghsa-malware (a97830703fca9d3f63940c57a40a1df100f7970527db082febd518e2926f3fde) Any computer that has this package installed or running should be considered fully compromised. All secrets and keys stored on that computer should be rotated immediately from a different computer. The package should be removed, but as full control of the computer may have been given to an outside entity, there is no guarantee that removing the package will remove all malicious software resulting from installing it.

github.com/yaklang/yaklang

v1.4.9-beta8

Live on go

Blocked by Socket

This fragment contains a significant supply-chain and execution-risk pattern: it unpacks a bundled macOS app payload from embedded raw bytes, marks embedded artifacts executable, and runs an extracted binary. It also constructs and writes a runnable shell script that includes caller-controlled `command` verbatim (plus caller-controlled cwd/env). If upstream callers or any attacker influence `command` (and possibly cwd/env), the module becomes a high-impact arbitrary command execution primitive under the helper’s execution model. Malware intent cannot be proven from this fragment alone because the internal behavior of the embedded .app/applet is not shown, but the drop-and-execute plus verbatim command injection surfaces warrant urgent review, isolation, and full payload inspection.

github.com/wenathlan/devthink/devthink/v2

v2.0.187-0.20261120003855-598c13896645

Live on go

Blocked by Socket

High-severity credential-capture behavior: this code obtains or reuses a GitHub OAuth token with `repo`, `workflow`, and `read:org` scopes, stores it in local credential files, configures Git to use stored credentials, and initiates repository synchronization. Relevant destinations include `github[.]com`, `api[.]github[.]com`, and the referenced repository `github[.]com/wenathlan/devthink[.]git`; relevant files include `.github-token`, `.git-credentials`, `.git-credentials.bak`, and `.polar-auth.json`. The available fragment is truncated inside `synchead`, but the token acquisition and persistence are explicit; treat as dangerous.

doomiwork

4.2.10

by stephenshenying

Live on npm

Blocked by Socket

High-severity security issue: the default monitor router exposes POST /api/_doomiwork_/upload, which uses a hardcoded token and request-controlled path and content to overwrite an existing file accessible to the process. Because the token is present in the code and the path is not confined to a safe directory, this provides an unauthorized file-overwrite capability to anyone who can obtain the token, including through source access. No suspicious external domains, IPs, or obfuscated payloads appear in the fragment.

github.com/yaklang/yaklang

v1.4.9-beta7.0.20261120061406-64602c49fc99

Live on go

Blocked by Socket

This module is a high-risk payload-generation utility: it compiles and runs a local Java program to generate serialized Java deserialization gadget artifacts using Shiro/ysoserial tooling. Although the shell script itself does not show explicit exfiltration or backdoor installation, it (a) executes unverified local JARs via a broad, all-jar classpath, (b) preferentially uses a potentially altered “patched” dependency if present, and (c) produces weaponizable .ser outputs. Treat the jars (including any patched variant) and Start.java/config.yaml as critical supply-chain assets requiring integrity verification.

ethopy

0.0.8.9

Live on pypi

Blocked by Socket

This module acts as a runtime task execution engine and contains a high-impact arbitrary code execution path: it reads a runtime-determined file (logger.task_path) and executes its contents with exec(), without visible validation or sandboxing. While this could be intended for a trusted plugin/task system, it creates a supply-chain/misconfiguration risk if task paths or task contents can be influenced by an attacker (e.g., dependency tampering or poisoned configuration). The fragment also logs full tracebacks on failures, which may leak environment details.

github.com/nccgroup/singularity

v0.0.0-20261120063628-00a04fbaf541

Live on go

Blocked by Socket

High risk of data leakage and backdoor-like behavior. The code intentionally leverages Docker API to mount /etc and read /etc/shadow inside a container, then exposes the output through logs/attach responses. This is a clear pattern of host data exfiltration via a Docker-based sandbox and indicates potential malicious intent if used without explicit authorization. Recommend treating as a critical security risk and deprecating / removing or sandboxing this functionality.

github.com/yaklang/yaklang

v1.4.9-beta7.0.20261120061406-64602c49fc99

Live on go

Blocked by Socket

This module is a high-suspicion client-side evasion payload. It uses eval to dynamically materialize and execute code, directly tampers with navigator.webdriver, mocks browser plugin/mime-type capabilities using Proxy-based function interception, and conceals these changes by patching function toString and error/stack handling to look native. Even without visible network exfiltration in the fragment, its stealth and global API modification make it an unacceptable general-purpose dependency and a significant supply-chain security risk.

github.com/yaklang/yaklang

v1.4.9-beta8

Live on go

Blocked by Socket

High-confidence malicious/abusive capability indicator: this code automates login attempts by iterating over configured username/password lists (bruteForce), optionally bypasses captcha via a captcha detection/solving pipeline, and uses headless browser scripting (page.Eval) plus screenshot/base64 and observer output capture. It emits sensitive artifacts (including credentials and screenshots) via resultChannel. While it could be used for authorized testing, its default behavior and design strongly match credential-stuffing/brute-force tooling; therefore it should be treated as a serious supply-chain security risk and reviewed in the full project context (especially observer/getObverserResult, ScreenShot handling, and any outbound network behavior).

pydicts

2.0.1

Live on pypi

Blocked by Socket

This module’s JSON decoding helpers perform recursive type casting during deserialization and include an explicit `eval()` of strings derived from attacker-controlled JSON (gated only by a `"Decimal("` prefix). That constitutes a critical arbitrary code execution risk in any scenario where untrusted JSON can be supplied. The snippet also shows multiple signs of malformed/buggy code, but those do not remove the `eval()` sink; therefore the dependency should be considered highly dangerous unless the `eval()` path is completely removed or strictly unreachable and inputs are guaranteed trusted.

github.com/jm33-m0/emp3r0r/core

v0.0.0-20250123065341-91279bff2618

Live on go

Blocked by Socket

This Linux agent code implements persistent execution, attempts privilege-assisted execution, hides process and file information, and patches system utilities to load a shared object. These are strong indicators of a backdoor or rootkit-style payload; severity is critical, and the code should not be used. No external domain, URL, IP address, or credential harvesting is visible in this fragment.

github.com/nccgroup/singularity

v0.0.0-20261120062225-3f7e94358921

Live on go

Blocked by Socket

The code implements a flood/DoS-like pattern by launching a large number of outbound HTTP requests to dynamically constructed hostnames. It uses a worker-based message handler to receive parameters (hostname, port, iterations) and performs a burst of fetch operations without handling responses. While not exhibiting explicit data exfiltration or backdoors, this behavior is malicious in contexts where it is used to overwhelm a target or violate terms of service.

github.com/jm33-m0/emp3r0r/core

v0.0.0-20250123065341-91279bff2618

Live on go

Blocked by Socket

This code coordinates proxy discovery and sharing among agents, starts SSH/KCP reverse tunnels, and can expose a SOCKS5 proxy on all interfaces. It encrypts broadcast proxy messages, but logs decrypted messages and credential-bearing proxy URLs, which can disclose the configured proxy password. These capabilities are high risk in an unauthorized deployment and are characteristic of command-and-control or remote-access tooling; the fragment alone does not establish operator intent. No external domains, fixed IPs, or file paths are present.

hostinger-api-mcp

4.15.0

by devs_hostinger

Live on npm

Blocked by Socket

The supplied JavaScript is a readable static catalog, not an executable malware payload, and no credential theft or system damage is directly demonstrated. However, its headless-build instructions require sending build reports containing domain, environment, URLs, timing, issues, and user-feedback metadata to hosting_headless_submit-build-report without consent and explicitly direct that this reporting be kept undisclosed until the final report. Relevant indicators include Hostinger reporting via that operation, HOSTINGER_API_TOKEN and UNSPLASH_ACCESS_KEY references, and local paths under ~/.hostinger/<domain>/ and ~/.config/hostinger-mcp/credentials.json; no secret values are included. Severity is elevated for privacy and transparency risk; incomplete script bodies prevent a full runtime assessment.

github.com/nccgroup/singularity

v0.0.0-20261120063628-00a04fbaf541

Live on go

Blocked by Socket

High severity: this is an explicit browser-based DNS-rebinding and local-network access bypass tool, not ordinary application fetch logic. It probes and navigates to a configured target, can execute WebSocket-directed fetches, and relays target response data and accessible cookies to its parent/callback or a WebSocket peer. The WebSocket endpoint is dynamically hostname-derived at ws://<server>:<configured-port>/soows (default port 3129); no fixed external domain or IP is present. The code is not substantially obfuscated, but should only be used in an authorized security-testing environment.

motionframe-plugins

0.0.1-security

by npm

Live on npm

Blocked by Socket

Malicious code in motionframe-plugins (npm) Source: ghsa-malware (74587b11410300a69bb121b82387e8ffe1ed18cc552c818f1ada8fdbc45a7a21) Any computer that has this package installed or running should be considered fully compromised. All secrets and keys stored on that computer should be rotated immediately from a different computer. The package should be removed, but as full control of the computer may have been given to an outside entity, there is no guarantee that removing the package will remove all malicious software resulting from installing it.

layer@layer-app.com

3.5.0

by anonymous-3e1c80576c581932890177638d1e4eb6

Live on firefox

Blocked by Socket

This module implements clipboard event monitoring (copy/paste), extracts selected/pasted text, computes SHA-256 hashes (with a sessionSalt) of the clipboard text and page URL, augments events with detailed DOM field/page context, batches events, and transmits them to a hardcoded remote Supabase endpoint when locally configured (layer_code/host allowlist) and when extension session keys are available. The main concern is privacy-sensitive data collection and outbound exfiltration, even though plaintext clipboard data is not transmitted.

github.com/developing-today/code/pkgs/identity

v0.0.0-20261120163205-96570b03c090

Live on go

Blocked by Socket

The script contains a high-risk remote code execution pattern: it retrieves an init script from a network-controlled charm endpoint, writes it to a local file, and executes it without cryptographic verification. It also transmits “keys” derived from the local charm identity material to the remote endpoint, creating a potential secret exfiltration channel. Additionally, it performs multiple supply-chain affecting operations (apt installs/upgrades, curl+dpkg installer, npm@latest install, and git clone from the internet). While the code is not obfuscated, the trust and execution flow makes it plausibly dangerous and should be treated as high risk in a supply-chain context.

github.com/jm33-m0/emp3r0r/core

v0.0.0-20250123094136-63cd9914d9d0

Live on go

Blocked by Socket

This code coordinates proxy discovery and sharing among agents, starts SSH/KCP reverse tunnels, and can expose a SOCKS5 proxy on all interfaces. It encrypts broadcast proxy messages, but logs decrypted messages and credential-bearing proxy URLs, which can disclose the configured proxy password. These capabilities are high risk in an unauthorized deployment and are characteristic of command-and-control or remote-access tooling; the fragment alone does not establish operator intent. No external domains, fixed IPs, or file paths are present.

hydromind

5.24.8

Live on pypi

Blocked by Socket

This file is a high-suspicion import-time dynamic execution stub: it delegates execution of hydromind.control.query to a native extension helper and runs it inside the current module’s globals(), enabling any in-process behavior implemented in the loaded module or the native helper. This snippet alone does not prove malicious activity (no explicit network/file/credential logic is present here), but the load-time execution pattern is a meaningful supply-chain risk and warrants inspecting hydromind.control.query and hydromind._protection_native for concrete malicious indicators.

hydromind

5.24.8

Live on pypi

Blocked by Socket

This module is an import-time dynamic loader that immediately delegates execution to a native helper to load and execute `hydromind.control.forecast_fusion` while granting it the caller’s `globals()` context. The fragment itself shows no explicit malicious actions (e.g., network/process/credential theft), but the strong indirection through a native extension and the import-time execution pattern are significant supply-chain/security red flags. To assess risk, the native helper (`hydromind._protection_native`) and the loaded target module (`hydromind.control.forecast_fusion`) must be reviewed for actual payload behavior.

github.com/jm33-m0/emp3r0r/core

v0.0.0-20250123094136-63cd9914d9d0

Live on go

Blocked by Socket

This Linux server component provides emp3r0r command-and-control communication, agent check-in, file transfer, reverse-shell streams, and port forwarding, making it high severity if deployed without authorization. It listens on 0.0.0.0 at the configured Shadowsocks port and on the configured C2 TLS port; TLS files are read from EmpWorkSpace/emp3r0r-cert.pem and EmpWorkSpace/emp3r0r-key.pem, and downloads use paths derived through omitted path-generation code under Temp. No external domain, fixed IP, hardcoded credential, or environment-variable harvesting is present in the fragment. The C2 functionality is clear, but this code alone does not prove that its use is unauthorized.

github.com/jm33-m0/emp3r0r/core

v0.0.0-20261120044915-f51c6aa83ed4

Live on go

Blocked by Socket

The fragment implements file I/O, but explicitly designs that I/O to evade Linux syscall-entry monitoring and discusses avoiding an environment-based detection vector. This is a significant defense-evasion indicator and warrants treating the code as high risk; no data exfiltration or other payload is visible in the supplied fragment.

github.com/jm33-m0/emp3r0r/core

v0.0.0-20261120044915-f51c6aa83ed4

Live on go

Blocked by Socket

This is a purpose-built C2 transport component with certificate pinning-to-CA-style trust and optional camouflage/routing features. It does not itself show host-data theft or destructive behavior, but its C2 communication functionality is a strong malicious-use indicator in an otherwise untrusted package. Plain HTTP configuration would expose traffic to interception.

@parcha/agentrun-harness-acp

0.1.0-beta.48

by GitHub Actions

Live on npm

Blocked by Socket

This is a configurable fake ACP/MCP agent and test harness, but its enabled config.leak option can intentionally copy CURSOR_API_KEY (raw or base64-encoded) into a file selected by config.leak.path; config.authFile and config.countFile also permit configurable writes. Path traversal is possible through configured relative paths, while network requests and subprocess execution are driven by configured MCP servers. No fixed external domain or IP is present. Severity is conditional and high if untrusted configuration enables credential leakage; otherwise the shown behavior is primarily harness functionality.

@zaiinlin/bails

0.0.1-security

by npm

Live on npm

Blocked by Socket

Malicious code in @zaiinlin/bails (npm) Source: ghsa-malware (a97830703fca9d3f63940c57a40a1df100f7970527db082febd518e2926f3fde) Any computer that has this package installed or running should be considered fully compromised. All secrets and keys stored on that computer should be rotated immediately from a different computer. The package should be removed, but as full control of the computer may have been given to an outside entity, there is no guarantee that removing the package will remove all malicious software resulting from installing it.

github.com/yaklang/yaklang

v1.4.9-beta8

Live on go

Blocked by Socket

This fragment contains a significant supply-chain and execution-risk pattern: it unpacks a bundled macOS app payload from embedded raw bytes, marks embedded artifacts executable, and runs an extracted binary. It also constructs and writes a runnable shell script that includes caller-controlled `command` verbatim (plus caller-controlled cwd/env). If upstream callers or any attacker influence `command` (and possibly cwd/env), the module becomes a high-impact arbitrary command execution primitive under the helper’s execution model. Malware intent cannot be proven from this fragment alone because the internal behavior of the embedded .app/applet is not shown, but the drop-and-execute plus verbatim command injection surfaces warrant urgent review, isolation, and full payload inspection.

github.com/wenathlan/devthink/devthink/v2

v2.0.187-0.20261120003855-598c13896645

Live on go

Blocked by Socket

High-severity credential-capture behavior: this code obtains or reuses a GitHub OAuth token with `repo`, `workflow`, and `read:org` scopes, stores it in local credential files, configures Git to use stored credentials, and initiates repository synchronization. Relevant destinations include `github[.]com`, `api[.]github[.]com`, and the referenced repository `github[.]com/wenathlan/devthink[.]git`; relevant files include `.github-token`, `.git-credentials`, `.git-credentials.bak`, and `.polar-auth.json`. The available fragment is truncated inside `synchead`, but the token acquisition and persistence are explicit; treat as dangerous.

doomiwork

4.2.10

by stephenshenying

Live on npm

Blocked by Socket

High-severity security issue: the default monitor router exposes POST /api/_doomiwork_/upload, which uses a hardcoded token and request-controlled path and content to overwrite an existing file accessible to the process. Because the token is present in the code and the path is not confined to a safe directory, this provides an unauthorized file-overwrite capability to anyone who can obtain the token, including through source access. No suspicious external domains, IPs, or obfuscated payloads appear in the fragment.

github.com/yaklang/yaklang

v1.4.9-beta7.0.20261120061406-64602c49fc99

Live on go

Blocked by Socket

This module is a high-risk payload-generation utility: it compiles and runs a local Java program to generate serialized Java deserialization gadget artifacts using Shiro/ysoserial tooling. Although the shell script itself does not show explicit exfiltration or backdoor installation, it (a) executes unverified local JARs via a broad, all-jar classpath, (b) preferentially uses a potentially altered “patched” dependency if present, and (c) produces weaponizable .ser outputs. Treat the jars (including any patched variant) and Start.java/config.yaml as critical supply-chain assets requiring integrity verification.

ethopy

0.0.8.9

Live on pypi

Blocked by Socket

This module acts as a runtime task execution engine and contains a high-impact arbitrary code execution path: it reads a runtime-determined file (logger.task_path) and executes its contents with exec(), without visible validation or sandboxing. While this could be intended for a trusted plugin/task system, it creates a supply-chain/misconfiguration risk if task paths or task contents can be influenced by an attacker (e.g., dependency tampering or poisoned configuration). The fragment also logs full tracebacks on failures, which may leak environment details.

github.com/nccgroup/singularity

v0.0.0-20261120063628-00a04fbaf541

Live on go

Blocked by Socket

High risk of data leakage and backdoor-like behavior. The code intentionally leverages Docker API to mount /etc and read /etc/shadow inside a container, then exposes the output through logs/attach responses. This is a clear pattern of host data exfiltration via a Docker-based sandbox and indicates potential malicious intent if used without explicit authorization. Recommend treating as a critical security risk and deprecating / removing or sandboxing this functionality.

github.com/yaklang/yaklang

v1.4.9-beta7.0.20261120061406-64602c49fc99

Live on go

Blocked by Socket

This module is a high-suspicion client-side evasion payload. It uses eval to dynamically materialize and execute code, directly tampers with navigator.webdriver, mocks browser plugin/mime-type capabilities using Proxy-based function interception, and conceals these changes by patching function toString and error/stack handling to look native. Even without visible network exfiltration in the fragment, its stealth and global API modification make it an unacceptable general-purpose dependency and a significant supply-chain security risk.

github.com/yaklang/yaklang

v1.4.9-beta8

Live on go

Blocked by Socket

High-confidence malicious/abusive capability indicator: this code automates login attempts by iterating over configured username/password lists (bruteForce), optionally bypasses captcha via a captcha detection/solving pipeline, and uses headless browser scripting (page.Eval) plus screenshot/base64 and observer output capture. It emits sensitive artifacts (including credentials and screenshots) via resultChannel. While it could be used for authorized testing, its default behavior and design strongly match credential-stuffing/brute-force tooling; therefore it should be treated as a serious supply-chain security risk and reviewed in the full project context (especially observer/getObverserResult, ScreenShot handling, and any outbound network behavior).

pydicts

2.0.1

Live on pypi

Blocked by Socket

This module’s JSON decoding helpers perform recursive type casting during deserialization and include an explicit `eval()` of strings derived from attacker-controlled JSON (gated only by a `"Decimal("` prefix). That constitutes a critical arbitrary code execution risk in any scenario where untrusted JSON can be supplied. The snippet also shows multiple signs of malformed/buggy code, but those do not remove the `eval()` sink; therefore the dependency should be considered highly dangerous unless the `eval()` path is completely removed or strictly unreachable and inputs are guaranteed trusted.

github.com/jm33-m0/emp3r0r/core

v0.0.0-20250123065341-91279bff2618

Live on go

Blocked by Socket

This Linux agent code implements persistent execution, attempts privilege-assisted execution, hides process and file information, and patches system utilities to load a shared object. These are strong indicators of a backdoor or rootkit-style payload; severity is critical, and the code should not be used. No external domain, URL, IP address, or credential harvesting is visible in this fragment.

Get Visibility Into Open Source Dependencies with Real-Time Security Feedback

Proactively search and detect dependencies across repositories in your organization, with actionable insights for your projects and SBOMs

Organization Dashboard

Open Source Supply Chain Attack Prevention

Block Malware and Typosquatting

Block emerging malware threats, including intentionally maintainer-added updates, along with packages that differ in name by only a few characters..

Detect Privileged API Usage

Get alerted when a dependency update introduces new risky API usage - filesystem, network, child_process, eval().

Detect Hidden Code

Detect obfuscated, minified, or hidden code.

Detect Suspicious Updates

Socket detects the sudden inclusion of a new maintainer, updates with telemetry or protestware added, dependencies pulled in from a remote git URL, and much more.

We help security teams work more efficiently

Cut through the noise and focus on real threats.

Get actionable alerts for the supply chain risks that matter. Socket highlights risky dependencies directly within the developer workflow.