Last Updated: 2026-05-14

pydepgate is a security tool, which means the bar for its own security is higher than for ordinary software. A vulnerability in pydepgate is worse than a vulnerability in most projects of comparable size because the tool is run against potentially malicious input by design.

This document describes how to report vulnerabilities in pydepgate, what counts as a vulnerability, and what to expect after reporting.

The preferred channel is GitHub's private vulnerability reporting:

1. Go to https://github.com/nuclear-treestump/pydepgate/security
2. Click "Report a vulnerability"
3. Fill out the form

This creates a private thread visible only to you and the maintainers. GitHub will help coordinate disclosure and CVE assignment if applicable.

If you prefer email, you can reach the maintainer at ikari@nuclear-treestump.com. Email reports may take longer to acknowledge than GHSA reports.

Please, do NOT file security issues on the public issue tracker.

If you require E2EE communication, please contact me via email so I can coordinate with you. I have Signal and PGP.

Vulnerabilities in pydepgate itself, including:

• Code execution in the pydepgate process triggered by scanning a crafted artifact. The static analyzer is designed never to execute user input; any path that does is a vulnerability.
• Path traversal or arbitrary file write during wheel or sdist extraction.
• Resource exhaustion (memory, CPU, file descriptors) triggered by a small crafted artifact, where the resource use is disproportionate to the input size.
• Output injection: a finding's content escaping the JSON or human reporter and producing misleading or malicious output.
• Credential or secret leakage through error messages, logs, or output formats.
• Vulnerabilities in pydepgate's installation path that would let an attacker tamper with the tool itself.

These are bugs, not security vulnerabilities. File them on the public issue tracker using the bug report template:

• pydepgate failed to detect a known piece of malware. The threat model is that pydepgate does not itself become a vector, not that it detects every possible attack. Detection gaps are bugs and feature requests.
• pydepgate flagged a benign artifact as malicious. False positives are bugs.
• pydepgate produced wrong severity, wrong line numbers, or other incorrect output for a non-security-impacting reason.
• Documentation errors, even ones that could mislead a user about what pydepgate detects.
• Third party interactions with pydepgate. I cannot anticipate how other software will interact with pydepgate. I've done my best to adhere to a standard, but if you find interesting interactions, please raise it.
• Purely AI sourced reports without user input. I am one person, not a corporation and I don't have unlimited time. Please respect my time and I will respect yours.
  • This is NOT meant to say "No AI assistance", just please verify your findings first.
• Malicious detections in test fixtures or tests/ directory. The tests are benign but look scary for the tool to test detection without detonating malware on my only computer.
• Cryptographic weaknesses in the encrypted-archive feature beyond what's already documented. The README states ZipCrypto is broken by design (AV-friendliness, not confidentiality); a vulnerability in scope here would be one that lets an attacker tamper with archive contents in a way the documented threat model doesn't account for, or that leaks plaintext under conditions that contradict the documentation.

If you're unsure whether something is in scope, err on the side of reporting it through the security channel. We would rather review a bug report through the security path than miss a real vulnerability because someone wasn't sure.

These are items I would prefer raised as feature requests or PRs. The Vulnerability Report workflow is the wrong tool for the job here. Please raise an Issue.

• New detection methods for pydepgate.
• Expanded documentation on the existing API.
• New operational modes.
• Other ideas that I haven't thought of.

Even if its just a conceptual idea, I will do my best to analyze and see if it is possible.

You will get credit if I make the improvement, unless you request otherwise.

1. Version of pydepgate in use

   pydepgate --version

2. OS in use
3. Version of Python in use

   Only 3.11, 3.12, 3.13, and beyond are in scope

4. Working proof of concept (a reproducer artifact, a command line, and the observed output)

   Reports that consist primarily of AI-generated analysis without a verified reproducer will be closed with a request for verification.

5. Explanation of how this vulnerability could be used or if applicable potential killchains.
6. Please confirm if you'd like credit or prefer to be anonymous.
7. (optional) If you have found a way to fix this, please share the fix so I can validate it. This is not required, but greatly appreciated.

pydepgate is currently maintained by one person on a side-project schedule. I check my email daily. The commitments below reflect that reality.

• Acknowledgment. I will acknowledge receipt within five business days. If you don't hear back in that window, the report didn't get to me, please send a follow-up.
  • The five days is the ceiling, not the floor. I will do my best to get back to you as fast as I am able.
• Triage. I will tell you whether the report is in scope and approximately what severity I'm assigning within two weeks.
• Fix. Timeline depends on severity:
  • Critical (remote code execution, sandbox escape): emergency release, target one week.
  • High: next scheduled release, target one month.
  • Medium and low: prioritized in the normal backlog.
• Disclosure. I will coordinate disclosure with you. The default is to publish a security advisory after a fix is released, crediting the reporter unless they prefer to remain anonymous.
• CVEs. CVEs will be requested through GitHub's CNA process. I haven't had to use that path yet, but the workflow is documented and I'll execute it when needed.
• Lack of Response. If I have not responded with a fix or a coordinated extension within 90 days, the reporter is free to disclose publicly.
  • Please reach out to me before it gets to this point. I do not want you to feel stuck and I don't want to blindside my users.

Reporters who follow this process and find legitimate vulnerabilities will be credited in the GitHub Security Advisory and the release notes for the version that contains the fix, unless they prefer otherwise.

pydepgate is a side project and does not offer monetary bug bounties. I do not have the money to pay you and currently build radiators for a living.

If you ARE a corporation looking for a security engineer, I am available.

If my living situation changes, a paid bug bounty may be enacted.

I will not pursue legal action against researchers who follow this policy in good faith, including testing against artifacts you control and reporting through the channels described in this document.

If you discover something so severe that even private GHSA reporting feels too slow (active exploitation in the wild, etc.), email ikari@nuclear-treestump.com with BUGCHECK (or GURUMEDITATION, if you prefer) in the subject. I check email more often than GitHub notifications and will respond faster.

pydepgate has no third-party runtime dependencies by design. If a vulnerability is reported against the Python standard library or CPython itself, that's not a pydepgate issue, but if pydepgate's use of the affected stdlib feature has security implications, that is in scope and we want to know.

If you're testing for vulnerabilities, please do so against your own local copy of pydepgate. Do not test against any service or infrastructure that hosts pydepgate (PyPI, GitHub Actions, etc.). Local testing on artifacts you control is welcome and encouraged.

v1.1 - 2026-05-03

What changed:

• Added ZipCrypto to out of scope bucket

v1.2 - 2026-05-12

What changed:

• Migration of reporting link from old name pydep-vector-runner to pydepgate

v1.3 = 2026-05-14

What changed:

• Updated missed changelog entry for v1.2, moved changelog to bottom of file.