Skip to content

Security: Nainish-Rai/agent-rollback

Security

SECURITY.md

Security Policy

Supported versions

agent-rollback is currently in active development. Security updates are released for the latest minor version.

Version Supported
1.0.x
< 1.0

Reporting a vulnerability

Please do not file a public GitHub issue for security vulnerabilities. Email nraibackup1@gmail.com (or use GitHub's private vulnerability reporting) instead.

Include:

  • A description of the vulnerability
  • Steps to reproduce
  • The impact (what an attacker could do)
  • Any known mitigations

We aim to acknowledge reports within 3 business days and ship a fix within 14 days for high-severity issues.

Threat model

agent-rollback is a local-only CLI and MCP server. It:

  • Reads and writes files only inside the workspace you point it at (or inside .agent-rollback/ by default).
  • Does not open network sockets, except for the optional npm install the installer runs.
  • Does not collect or send telemetry.
  • Runs the Codex hook scripts with the same privileges as the Codex CLI process.

Things agent-rollback does not protect against:

  • Malicious Codex output that has been approved by the user
  • Supply-chain attacks against npm dependencies (@modelcontextprotocol/sdk, zod)
  • Privilege escalation in the workspace

Best practices for users

  • Keep Node.js up to date (we require >= 20).
  • Use npm audit to check for known issues in dependencies.
  • Pin the version when installing for production: npm install -g agent-rollback@1.0.1.
  • Review Codex's diffs before approving any file write.
  • Back up your .agent-rollback/ directory if it contains valuable checkpoints (it is yours, not your repo's).

Disclosure policy

We follow coordinated disclosure. Please give us a reasonable window to fix before public disclosure.

There aren't any published security advisories