Skip to content

Security: Abhinav-Rust/REFAL-SUPERSYSTEM

Security

SECURITY.md

Security policy

Supported versions

The project is developed on main, and security fixes land on main and in the next release. There is no long-term-support branch.

Reporting a vulnerability

Please do not open a public issue for a security problem. Use GitHub's private vulnerability reporting: Security → Advisories → Report a vulnerability.

Please include:

  • what the defect is, and the command or input that triggers it;
  • the affected version or commit (git rev-parse --short HEAD);
  • the impact you believe it has;
  • any suggested fix, if you have one.

You can expect an acknowledgement within a few days, and then either an assessment with a remediation plan or a reasoned decision that the report is out of scope.

Scope

This is a compiler and a program-transformation engine, so the interesting surface is what it does with input it is given:

  • the lexer, parser and semantic checker, driven by an untrusted .ref source;
  • the Refal machine, driven by an untrusted program;
  • the driver and supercompiler, whose termination and resource use are bounded by an explicit step budget.

In scope:

  • a defect that lets an input escape the process it runs in — a panic reachable from a valid program, an unbounded allocation or loop where a budget is promised, a file read or write outside the path it was given;
  • a wrong program — a residue or a compiled artifact that does not agree with the source it came from. This project treats that as a correctness defect of the first order, and it is the one class of bug its gates exist to catch.

Out of scope: denial of service by handing the tool an intentionally enormous input, and anything that requires an attacker to already control the machine.

How the project already hardens itself

  • Secret scanning and push protection are enabled for this repository.
  • Dependabot watches the dependency graph and opens security updates.
  • Every push to main runs cargo fmt --check, cargo clippy --all-targets -- -D warnings, and the full corpus gates.

There aren't any published security advisories