Malicious code in seekcode (npm)
The seekcode npm package contains malicious code that redirects users selecting the deepseek-cn provider to a typosquatted domain (api.deepseeki.com instead of api.deepseek.com), exfiltrating API credentials and chat prompt contents to an attacker-controlled server.
- Disclosed
- Last updated
- Blast radius
- All users of the seekcode npm package who select the deepseek-cn provider will have their DeepSeek API credentials and chat prompt contents exfiltrated to an attacker-controlled domain.
- Ecosystems
- Attack vectors
- Affected entities
- seekcodenpm package containing malicious code in the defaultBaseUrlForProvider function
The seekcode npm package was found to contain intentional malicious code in its defaultBaseUrlForProvider function (located in dist/chunk-6U42R724.js). When users select the advertised deepseek-cn provider, the function returns https://api.deepseeki.com — a one-character typosquat of the legitimate DeepSeek endpoint api.deepseek.com.\n\nAll other provider branches in the same switch statement correctly return their official endpoints (api.deepseek.com, integrate.api.nvidia.com, openrouter.ai, etc.), making this deviation intentional rather than accidental. The asymmetry strongly indicates deliberate malicious intent: a random typo would be unlikely to consistently land on a lookalike domain while all sibling cases remain correct.\n\nAny user invoking the deepseek-cn provider will send their DeepSeek API bearer token and the full content of every chat prompt to the attacker-controlled domain. This represents a critical credential and data exfiltration risk, as chat prompts routinely contain private code, secrets, and proprietary data when used through coding assistants.\n\nThe incident was identified and credited to the OpenSSF's malicious-packages project.
Indicators of compromise
- Domains
- api.deepseeki.com
Remediation
- Immediately remove or update the seekcode package to a patched version that corrects the deepseek-cn provider endpoint to the legitimate api.deepseek.com
- Audit npm package dependencies for seekcode and remove it if not actively maintained or if no patched version is available
- If seekcode was used with the deepseek-cn provider, rotate all DeepSeek API credentials immediately
- Review chat history and prompts sent through seekcode for any sensitive data exposure
- Monitor the attacker-controlled domain (api.deepseeki.com) for evidence of credential or data exfiltration
Sources
- GitHub Advisory GHSA-xj6w-x4p9-6r73 · GitHub Advisory Database
Cite this entry
"Malicious code in seekcode (npm)." supplychainattack.org, Supply Chain Attack Incident Catalog. Disclosed July 27, 2026; last updated July 27, 2026. https://supplychainattack.org/incident/malicious-code-in-seekcode-npm-buk7v1
Suggest a correction
Found an error or have a newer source? Corrections to factual errors take priority over new entries.
Related incidents
- containedcritical
Malicious code in @coralxyz/anchor (npm)
The npm package @coralxyz/anchor is a typosquatting attack impersonating the legitimate @coral-xyz/anchor Solana framework. It contains a postinstall hook that downloads and executes a malicious executable on Windows systems. The package provides no legitimate functionality and serves solely as a dropper for malware.
npmTyposquattingCompromised package - resolvedcritical
Malicious code in dolyame-ui-postcsscustomproperties (npm)
The npm package dolyame-ui-postcsscustomproperties contained malicious code that downloads and executes platform-specific binaries from attacker-controlled Cloudflare Workers and DNS infrastructure. The package typosquats the legitimate postcss-custom-properties package and disguises the malicious payload as an analytics/telemetry module.
npmCompromised packageTyposquatting - containedcritical
Malicious code in base-ui-cli (npm)
The npm package base-ui-cli contains malicious code that makes unauthorized fetch() calls to an attacker-controlled Cloudflare Workers endpoint (base-ui-pro-registry.l-dimitrov.workers.dev) in addition to the legitimate npm registry. The package name appears designed to impersonate the unrelated Base UI component library, suggesting a typosquatting attack.
npmCompromised packageTyposquatting - activecritical
Malicious code in hardhat-set (npm)
The npm package hardhat-set contains malicious obfuscated code disguised with a copied pino logger README. The package executes a multi-megabyte obfuscated payload on import, delivering hidden malware to anyone who installs or requires it.
npmTyposquattingCompromised package