Skip to content

Security: EffortlessMetrics/perl-lsp

Security

SECURITY.md

Security Policy

Supported Versions

Version Supported
0.11.x Yes -- current release line
0.10.x Security fixes only
< 0.10 No

Reporting a Vulnerability

Please do not open public issues for security vulnerabilities.

We prefer reports through GitHub Security Advisories. If that is not possible, email the maintainers listed in the root Cargo.toml.

What to include

  • Description of the vulnerability
  • Steps to reproduce
  • Potential impact assessment
  • Suggested fix, if you have one

Response timeline

Milestone Target
Acknowledge report 48 hours
Severity assessment 72 hours
Fix for critical/high 30 days
Coordinated disclosure After fix is released

We follow coordinated disclosure. You will be notified before any public advisory, and we will credit you unless you prefer to remain anonymous.

Scope

The following are considered security issues for this project:

  • Path traversal in file operations (e.g., workspace file access escaping configured roots)
  • Arbitrary code execution triggered by LSP protocol messages or parsed input
  • LSP protocol injection (malformed messages causing unintended server behavior)
  • Denial of service through crafted input that causes unbounded resource consumption
  • Dependency vulnerabilities in crates shipped as part of perl-lsp

Out of scope

  • Bugs in the Perl source code being analyzed (perl-lsp is a read-only tool; it does not execute Perl)
  • Editor or client-side issues (report those to the editor/extension maintainer)
  • Vulnerabilities that require local shell access beyond what the LSP client already provides

Security Practices

  • All production code is written in safe Rust. Fatal constructs (unwrap, panic!, process::abort) are banned outside of tests.
  • Dependencies are audited with cargo-audit and cargo-deny; see deny.toml for policy.
  • Fuzz testing covers the parser and lexer.
  • LSP communication uses stdio by default (no network listener).
  • File system access is limited to workspace roots configured by the client.

Last updated: 2026-03-11

There aren't any published security advisories