Skip to content

Security: Mullassery/PyBeamGuard

Security

SECURITY.md

Security Policy

Supported versions

PyBeamGuard is a single-maintainer, pre-1.0-maturity-in-spirit project (currently at 1.2.1). There is no formal LTS or backport policy — only the latest released version on PyPI / GitHub Releases receives fixes.

Version Supported
1.2.x (latest) Yes
< 1.2.0 No — please upgrade

Reporting a vulnerability

Please report security issues privately rather than opening a public GitHub issue:

Include:

  • A description of the issue and its potential impact.
  • Steps to reproduce, or a minimal proof-of-concept pipeline/input.
  • The PyBeamGuard version and installation method (PyPI wheel, standalone binary, or built from source).

There is no bug bounty program. As a single-maintainer project, response time is best-effort — expect an initial acknowledgment within a week, not guaranteed same-day.

What counts as a security issue here

PyBeamGuard is a static analysis CLI that reads local pipeline source files (and optional local JSON/YAML config files) and prints a report. It does not execute the pipeline code it analyzes, make network calls, or run with elevated privileges. Realistic security-relevant classes for this project are:

  • A crafted pipeline source file, --data-profile JSON file, or --rules YAML file that causes a crash, panic, hang, or resource exhaustion (denial of service) rather than a clean error. The Rust core has 114+ .unwrap()/.expect() call sites (see ROADMAP_HONEST.md) that have not been systematically audited for panic-on-malformed-input; the integration test suite covers some malformed-input cases (crates/core/tests/integration.rs: empty source, non-Beam source, malformed data-profile JSON, pathological Unicode) but this is not a claim of full coverage.
  • A memory-safety issue in the Rust core or the PyO3 bindings layer (bindings/python/src/lib.rs) reachable from untrusted input.
  • Supply-chain issues in dependencies — see cargo audit results in CI (.github/workflows/ci.yml, security-audit job, informational/ non-blocking as of this writing — see ROADMAP_HONEST.md).

This tool does not process secrets, credentials, or PII by design, and has no telemetry or network calls of its own.

Known gaps

See the "Technical Debt" section of ROADMAP_HONEST.md for the current, honest list of unaudited panic surface and dependency-scanning gaps — these are documented there rather than silently left undocumented.

There aren't any published security advisories