Vulnerability-Lookup

PYSEC-2026-1075

Vulnerability from pysec - Published: 2026-07-07 16:26 - Updated: 2026-07-07 16:26

Details

Part of the "Hades" wave of the Shai-Hulud supply-chain campaign. On 2026-06-08, malicious phantom releases of tiktoken-mcp were published to PyPI using stolen credentials. The package executes a bundled JavaScript payload (via the Bun runtime) on import that harvests and exfiltrates credentials and attempts self-propagation. This entry is a summary; behavior may not be fully characterized here. See the linked references for detailed analysis and indicators of compromise.

Impacted products

Name	purl
tiktoken-mcp	pkg:pypi/tiktoken-mcp

Aliases

MAL-2026-5326

JSON

To clipboard

{
  "affected": [
    {
      "package": {
        "ecosystem": "PyPI",
        "name": "tiktoken-mcp",
        "purl": "pkg:pypi/tiktoken-mcp"
      },
      "versions": [
        "0.13.1",
        "0.13.2"
      ]
    }
  ],
  "aliases": [
    "MAL-2026-5326"
  ],
  "details": "Part of the \"Hades\" wave of the Shai-Hulud supply-chain campaign. On 2026-06-08,\nmalicious phantom releases of tiktoken-mcp were published to PyPI using stolen\ncredentials. The package executes a bundled JavaScript payload (via the Bun\nruntime) on import that harvests and exfiltrates credentials and attempts\nself-propagation. This entry is a summary; behavior may not be fully\ncharacterized here. See the linked references for detailed analysis and\nindicators of compromise.\n",
  "id": "PYSEC-2026-1075",
  "modified": "2026-07-07T16:26:46Z",
  "published": "2026-07-07T16:26:46Z",
  "references": [
    {
      "type": "EVIDENCE",
      "url": "https://inspector.pypi.io/project/tiktoken-mcp/0.13.2/packages/5c/58/895febe9c7cc34d3e4d64699ec40ac0b378392db3cfda3c93b7af27da95d/tiktoken_mcp-0.13.2-py3-none-any.whl/tiktoken-setup.pth"
    },
    {
      "type": "ARTICLE",
      "url": "https://www.endorlabs.com/learn/shai-hulud-hades-wave-hits-six-pypi-bioinformatics-packages"
    },
    {
      "type": "ARTICLE",
      "url": "https://www.stepsecurity.io/blog/the-hades-campaign-pypi-packages"
    }
  ],
  "summary": "Malicious code in tiktoken-mcp (PyPI)"
}

mal-2026-5326

Vulnerability from ossf_malicious_packages

Published

2026-06-08 10:16

Modified

2026-06-11 07:52

Summary

Malicious code in tiktoken-mcp (PyPI)

Details

-= Per source details. Do not edit below this line.=-

Source: amazon-inspector (ac746100211f13951c190e98140c6948be51d7be9257b2b26bcc9baef19be29f)

tiktoken-mcp impersonates the OpenAI-published tiktoken package: its METADATA copies the upstream Name/Summary, Author 'Shantanu Jain', Author-email 'shantanu@openai.com', and Project-URL pointing at github.com/openai/tiktoken, with the upstream README bundled. The package ships tiktoken-setup.pth, which Python's site.py auto-executes at every interpreter start. The.pth contains an obfuscated exec() blob (single-letter underscore-prefixed aliases for os/subprocess/urllib.request/platform/sys/shutil/glob) that, on first run, downloads the Bun JS runtime from https://github.com/oven-sh/bun/releases/download/bun-v1.3.13/bun--.zip, extracts it to /tmp/b/bun, then walks sys.path searching for any file named '_index.js' in any package directory and executes it with 'bun run'. The package itself does not ship an _index.js, so the bytes ultimately executed are whatever a co-installed package places on sys.path under that name — i.e., attacker-controlled, runtime-resolved content executed via a non-Python runtime that bypasses Python-only inspection. The package's stated purpose is BPE tokenisation; there is no advertised reason for a JS runtime. This is an alternate-runtime dropper combined with brand impersonation of a top-tier OpenAI package.

Source: kam193 (c9be15ab63daf09fd0949c09ea93f0d014aa6886b071ecc5a1af0dc4546d5a2a)

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

CWE

CWE-506 - The product contains code that appears to be malicious in nature.

Credits

Amazon Inspector inspector-research@amazon.com

Kamil Mańkowski (kam193) github.com/kam193 bad-packages.kam193.eu/

JSON

To clipboard

{
  "affected": [
    {
      "database_specific": {
        "cwes": [
          {
            "cweId": "CWE-506",
            "description": "The product contains code that appears to be malicious in nature.",
            "name": "Embedded Malicious Code"
          }
        ],
        "indicators": {
          "evidence_files": [
            {
              "path": "tiktoken-setup.pth",
              "sha256": "6506d31707a39949f89534bf9705bcf889f1ecae3dbc6f4ff88d67a8be3d01b2",
              "tlsh": "8721015780038160da72d627cb2929f4278b0cbb4e538b1b8de8d9c172d0d1197b6b48"
            },
            {
              "path": "tiktoken_mcp-0.13.1.dist-info/METADATA",
              "sha256": "94fb43588dc4cd8c755fc9c248e672fa9d54489995fa9f65f818e27d41c8a21e",
              "tlsh": "cad1291be30c67601b862eb0b16be9bdef3cb19d6b0a5e87392c93e00f4011980b7565"
            }
          ],
          "package_integrity": [
            {
              "filename": "tiktoken_mcp-0.13.1-cp310-cp310-macosx_10_12_x86_64.whl",
              "hashes": {
                "blake2b_256": "0f9a817780a25f005cf49b0311856e5514e9558749b63d9c7e6dc73625236e04",
                "md5": "0e39996d1770e97c3b996071b9a6cf9d",
                "sha256": "a60c3945431c0a8fe7fd99c0227f2d75b6acc14358237cef56ac15fc9ff0b574"
              }
            }
          ]
        }
      },
      "package": {
        "ecosystem": "PyPI",
        "name": "tiktoken-mcp"
      },
      "versions": [
        "0.13.1",
        "0.13.2"
      ]
    }
  ],
  "credits": [
    {
      "contact": [
        "inspector-research@amazon.com"
      ],
      "name": "Amazon Inspector",
      "type": "FINDER"
    },
    {
      "contact": [
        "https://github.com/kam193",
        "https://bad-packages.kam193.eu/"
      ],
      "name": "Kamil Ma\u0144kowski (kam193)",
      "type": "ANALYST"
    }
  ],
  "database_specific": {
    "malicious-packages-origins": [
      {
        "id": "pypi/2026-06-compr-woodpecker/tiktoken-mcp",
        "import_time": "2026-06-08T11:41:02.549263659Z",
        "modified_time": "2026-06-08T10:16:53.930162Z",
        "sha256": "c9be15ab63daf09fd0949c09ea93f0d014aa6886b071ecc5a1af0dc4546d5a2a",
        "source": "kam193",
        "versions": [
          "0.13.1",
          "0.13.2"
        ]
      },
      {
        "id": "pypi/2026-06-compr-woodpecker/tiktoken-mcp",
        "import_time": "2026-06-09T07:48:29.66155438Z",
        "modified_time": "2026-06-08T10:16:53.930162Z",
        "sha256": "05e1b139b928815987413cc4db186e23aceff0d04df4fabaf7c80813b9b154b4",
        "source": "kam193",
        "versions": [
          "0.13.1",
          "0.13.2"
        ]
      },
      {
        "id": "IN-MAL-2026-005617",
        "import_time": "2026-06-11T07:49:33.215742097Z",
        "modified_time": "2026-06-11T06:16:39Z",
        "sha256": "ac746100211f13951c190e98140c6948be51d7be9257b2b26bcc9baef19be29f",
        "source": "amazon-inspector",
        "versions": [
          "0.13.1"
        ]
      }
    ]
  },
  "details": "\n---\n_-= Per source details. Do not edit below this line.=-_\n\n## Source: amazon-inspector (ac746100211f13951c190e98140c6948be51d7be9257b2b26bcc9baef19be29f)\ntiktoken-mcp impersonates the OpenAI-published tiktoken package: its METADATA copies the upstream Name/Summary, Author \u0027Shantanu Jain\u0027, Author-email \u0027shantanu@openai.com\u0027, and Project-URL pointing at github.com/openai/tiktoken, with the upstream README bundled. The package ships tiktoken-setup.pth, which Python\u0027s site.py auto-executes at every interpreter start. The.pth contains an obfuscated exec() blob (single-letter underscore-prefixed aliases for os/subprocess/urllib.request/platform/sys/shutil/glob) that, on first run, downloads the Bun JS runtime from https://github.com/oven-sh/bun/releases/download/bun-v1.3.13/bun-\u003cplatform\u003e-\u003carch\u003e.zip, extracts it to /tmp/b/bun, then walks sys.path searching for any file named \u0027_index.js\u0027 in any package directory and executes it with \u0027bun run\u0027. The package itself does not ship an _index.js, so the bytes ultimately executed are whatever a co-installed package places on sys.path under that name \u2014 i.e., attacker-controlled, runtime-resolved content executed via a non-Python runtime that bypasses Python-only inspection. The package\u0027s stated purpose is BPE tokenisation; there is no advertised reason for a JS runtime. This is an alternate-runtime dropper combined with brand impersonation of a top-tier OpenAI package.\n\n## Source: kam193 (c9be15ab63daf09fd0949c09ea93f0d014aa6886b071ecc5a1af0dc4546d5a2a)\nTyposquatting 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\u0027s data if it detects invalid GitHub tokens. Cleanup should be done with caution.\n\nIt seems to be related to the recent Mini Shai Hulud campaign.\n\n\n---\n\nCategory: MALICIOUS - The campaign has clearly malicious intent, like infostealers.\n\n\nCampaign: 2026-06-compr-woodpecker\n\n\nReasons (based on the campaign):\n\n\n - compromised-package\n\n\n - exfiltration-env-variables\n\n\n - exfiltration-cloud-tokens\n\n\n - exfiltration-credentials\n\n\n - abuses-pth\n\n\n - obfuscation\n\n\n - infostealer\n\n\n - The package contains code to detect if it is running in a sandbox environment.\n\n\n - exfiltration-crypto\n\n\n - files-exfiltration\n\n\n - destructive-actions\n",
  "id": "MAL-2026-5326",
  "modified": "2026-06-11T07:52:13Z",
  "published": "2026-06-08T10:16:53Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://socket.dev/blog/shai-hulud-descends-to-hades-miasma-pypi-wave"
    },
    {
      "type": "WEB",
      "url": "https://o3.security/blog/pypi-supply-chain-attack-pth-file-miasma"
    },
    {
      "type": "WEB",
      "url": "https://bad-packages.kam193.eu/pypi/package/tiktoken-mcp"
    },
    {
      "type": "WEB",
      "url": "https://socket.dev/blog/mini-shai-hulud-miasma-and-hades-worms-target-bioinformatics-and-mcp-developers-via-malicious"
    },
    {
      "type": "PACKAGE",
      "url": "https://pypi.org/project/tiktoken-mcp/0.13.1/"
    }
  ],
  "schema_version": "1.7.4",
  "summary": "Malicious code in tiktoken-mcp (PyPI)"
}

Sightings

Author	Source	Type	Date	Other

Nomenclature

Seen: The vulnerability was mentioned, discussed, or observed by the user.
Confirmed: The vulnerability has been validated from an analyst's perspective.
Published Proof of Concept: A public proof of concept is available for this vulnerability.
Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
Not confirmed: The user expressed doubt about the validity of the vulnerability.
Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.

Detection rules are retrieved from Rulezet.

Action not permitted