openai-mcp (PyPI) · Malicious Code
Affects openai-mcpSource GCVE Database
Malicious package : remove openai-mcp and rotate any secrets on hosts that installed it.
Data as of 11 Jul 2026, 09:43 UTC
Malicious package
Published to PyPI · OSV MAL-2026-5320
What to do
Remove openai-mcp from every project, lockfile and image that pulls it in. Treat any host that installed it as compromised: rotate the credentials, tokens and keys it could read, from a clean machine.
Affected software
| Product | Affected | Fixed in |
|---|---|---|
| PyPIopenai-mcp | Affected
| Fixed in– |
Check your own hosts for these packages with threat-finder (opens in a new tab), our free open-source scanner.
Description
--- _-= Per source details. Do not edit below this line.=-_ ## Source: amazon-inspector (6c30f8b1a160c72529e2671b81be13ea671302e9bc3915ee84bd2212e0fdd5a3) The package name and module structure impersonate the official `openai` Python SDK (the package even rewrites exported classes' `__module__` attribute to `"openai"` in `openai_mcp/__init__.py`). Alongside the cloned SDK, it ships `openai-setup.pth`, which Python auto-loads on every interpreter start. The.pth file contains a single-line, heavily aliased `exec()` payload (single-letter names like `_O`, `_T`, `_u`, `_s`, `_b`, `_z`) that: (1) checks a one-shot guard `/tmp/.bun_ran`, (2) downloads the Bun JavaScript runtime from `https://github.com/oven-sh/bun/releases/download/bun-v1.3.13/bun-<platform>-<arch>.zip` to `/tmp/b/bun` and chmods it executable, (3) locates a sibling `_index.js` on `sys.path`, and (4) runs `bun run _index.js`. Because.pth files execute at every Python startup (not just at install), every `python` invocation after `pip install openai-mcp` downloads an alien runtime and executes the package's JavaScript payload through it — an alternate-runtime dropper that evades Python-only scanners. The obfuscation, one-shot guard, identity spoofing, and runtime-laundering together establish clear malicious intent. ## Source: kam193 (f7597f40e5a7af25ae5f07e34310df24ac3284b9e0cadcb013a9c155384e9100) Typosquatting package published from a compromised account with an embedded infostealer. The infostealer is a heavily obfuscated JavaScript code executed using Bun runtime on Python startup. It collects all kinds of sensitive data, including API keys, credentials to package repositories, cryptocurrency assets, password manager data. Infostealer actively queries online services to collect additional secrets as well as attempts to gain persistence and spread further by publishing infected packages using collected credentials. Data are exfiltrated likely using Github. The code seems to threaten to wipe the user's data if it detects invalid GitHub tokens. Cleanup should be done with caution. It seems to be related to the recent Mini Shai Hulud campaign. --- Category: MALICIOUS - The campaign has clearly malicious intent, like infostealers. Campaign: 2026-06-compr-woodpecker Reasons (based on the campaign): - compromised-package - exfiltration-env-variables - exfiltration-cloud-tokens - exfiltration-credentials - abuses-pth - obfuscation - infostealer - The package contains code to detect if it is running in a sandbox environment. - exfiltration-crypto - files-exfiltration - destructive-actions
References
- socket.dev/blog/shai-hulud-descends-to-hades-miasma-pypi-wave (opens in a new tab)
- o3.security/blog/pypi-supply-chain-attack-pth-file-miasma (opens in a new tab)
- bad-packages.kam193.eu/pypi/package/openai-mcp (opens in a new tab)
- socket.dev/blog/mini-shai-hulud-miasma-and-hades-worms-target-bioinformatics-and-mcp-developers-via-malicious (opens in a new tab)
- pypi.org/project/openai-mcp/2.41.1/ (opens in a new tab)
- pypi.org/project/openai-mcp/2.41.2/ (opens in a new tab)