Skip to content

Security: Amz34/selfhosted-agent-stack

Security

SECURITY.md

Security Policy

Supported versions

This repository ships scripts and documentation, not a hosted service. Fixes land on the main branch; use the latest commit or the newest tagged release.

Reporting a vulnerability

Do not open a public issue. Report privately via GitHub's Report a vulnerability form (Security → Advisories → Report a vulnerability).

Please include:

  • affected file(s) and version/commit,
  • a minimal reproduction (command + config shape, with secrets redacted),
  • the impact you believe it has (privilege escalation, data exposure, denial of service),
  • any suggested mitigation.

You can expect an acknowledgement within a few days. Credit is given in the release notes unless you prefer to stay anonymous.

Scope notes for operators

These scripts run on your infrastructure with your privileges. Before deploying:

  • Review every script before running it, especially anything executed with sudo.
  • Never commit .env files or tokens. Load secrets from environment variables or a secret manager; keep files at mode 600.
  • Keep one credential scope per component. A bot token, API key or OAuth client used by an agent should be dedicated to that agent — shared tokens turn a single compromise into a full-stack one, and produce hard-to-debug runtime conflicts.
  • Rotate on suspicion, not on schedule — but do rotate after any suspected leak; a secret deleted from a working tree is still recoverable from git history.
  • Treat inbound webhooks as hostile input. Validate signatures, bound payload sizes, and never pass webhook content straight into a shell.

No script in this repository phones home, collects telemetry, or contacts any host other than the ones you explicitly configure.

There aren't any published security advisories