Python (CPython)

Skip to content What this pays This project runs no bounty of its own. sorona.xyz pays you directly on qualified, reproduced findings. Severity │ sorona.xyz pays Critical │ $1,500+ High │ $700+ Medium │ $300+ Low │ $100 / 1 month Claude Max Submit a vulnerability Separately, for a merged upstream patch : $100 - $5,000 Google Patch Rewards pays this project as a Tier 2 target (it qualifies through its OSS-Fuzz integration), for a patch you land upstream and that stays merged, without reverts, for one month. It is paid for the patch, not for the report, and it is rate-limited to three submissions a month. Sanctions and territory exclusions apply. CVE history 14 Last 12 months 148 Lifetime Memory safety 35 % Denial of service 22 % Parsing/validation 18 % Injection 13 % Latest advisories Loading the feed Overview CPython is the reference implementation of the Python language: the interpreter that runs Python code plus the batteries-included standard library that ships with it. Security reports are handled by the Python Security Response Team (PSRT), a small, trusted group of core developers who triage findings privately and coordinate fixes and CVEs. This project runs a vulnerability disclosure program with no bounty of its own. Report through sorona.xyz and we pay you directly on qualified, reproduced findings, with payouts scaling up to 2x the base for hard, chained bugs. In scope The CPython interpreter core: the eval loop, bytecode compiler, memory management, garbage collector, and object model, where a real-world exploitable flaw exists. Standard-library modules that parse or handle untrusted input: http, http.client, http.server, urllib, urllib.parse, email, ssl, ftplib, and smtplib. XML parsers (xml.sax, xml.dom, xml.etree.ElementTree, xml.parsers.expat) for XXE, entity-expansion, and billion-laughs style bugs. Archive and compression handling (zipfile, tarfile, gzip, bz2, lzma) for path traversal, symlink extraction, and decompression bombs, plus json, cgi, wsgiref, ipaddress, hashlib, and hmac. C-level accelerator and extension modules shipped with CPython (_ssl, _hashlib, _socket, _pickle, _json, _elementtree/expat, _ctypes, _sqlite3, and the zlib bindings) for memory-safety defects. String, bytes, and codec handling, plus path handling (os.path, pathlib) and subprocess argument or quoting handling where it enables injection beyond documented behavior. pip as distributed via PyPI, which the Python Security Response Team accepts alongside CPython. Platform-specific issues only where the platform meets a PEP 11 support tier. Out of scope Documented dangerous APIs used as designed: pickle, marshal, shelve, eval, exec, and ctypes, all documented to execute or expose arbitrary code or memory. Bugs reachable only through code, configurations, or deployments unlikely to occur in a real production program. Severity is scored from expected use, not the worst case. Unhandled exceptions treated on their own as an expected part of control flow. Flaws in bundled third-party dependencies such as zlib, OpenSSL, Tcl/Tk, or libexpat, unless CPython's own usage undermines their safe use. Experimental or provisional features, even in a stable release, and pre-release (alpha, beta, RC) or in-development code. Python versions past end-of-life, since only releases in bugfix or security status are accepted. Third-party redistributions of Python or pip and PyPI package vulnerabilities, plus issues that require the attacker to already control launch arguments, environment variables, or on-disk files. Most-valued findings Memory-safety bugs in the interpreter and C extensions: buffer overflows, out-of-bounds read or write, use-after-free, and integer overflow that corrupts memory. Remote or arbitrary code execution reachable from untrusted input through an API documented as safe. Path traversal and arbitrary file write or overwrite via zip/tar extraction or predictable tempfiles, and XXE or entity-expansion denial of service in the XML parsers. TLS or SSL certificate and hostname-verification bypasses and weak default crypto in the ssl module. Injection reachable beyond documented behavior: HTTP header or CRLF request injection, SQL through the sqlite bindings, and command or argument injection via subprocess or shutil. Algorithmic-complexity or decompression-bomb denial of service triggerable with reasonably sized input, plus information disclosure across a trust boundary. Program rules Submit through sorona.xyz. This project runs no bounty of its own; we validate, coordinate disclosure with the maintainers, and pay you directly. Report privately first and give the PSRT time to fix before any public discussion Include a minimal proof-of-concept with clear pass or fail indicators and the versions tested Show a realistic, exploitable path rather than a theoretical or purely lateral advantage Keep availability findings to reasonably sized inputs that a normal deployment would accept Submit reports in plain text and, where possible, include a minimal candidate patch Do not access, modify, or exfiltrate data belonging to other users or systems Disclosure Coordinated disclosure applies: the PSRT triages the report, assigns a CVE through the PSF CNA once accepted, and publishes an advisory with attribution only after a fix lands publicly, targeting the industry-standard 90-day window. Please hold public details until the advisory is out. Safe harbor Sorona will not pursue legal action against you for good-faith research submitted to us in line with this policy. We cannot speak for the upstream project, its vendor, or any third party, and we do not grant authorization on their behalf. You are responsible for understanding and following the laws that apply to you, wherever you are, and for staying within any scope the project has defined. Submit a vulnerability