Skip to content

Security: jordannn02/Codex-Second-Brain

Security

SECURITY.md

Security Policy

Codex Second Brain OS is a public-safe reference package for agent memory workflows. It should publish methods, schemas, fixtures, and redacted examples, not private memory.

Supported Scope

This repository contains:

  • public operating rules;
  • synthetic demo vault content;
  • schemas for route memory and capture events;
  • a minimal dry-run-first CLI;
  • tests for public fixtures.

It must not contain:

  • credentials, tokens, keys, cookies, or secrets;
  • raw private notes, transcripts, screenshots, or financial records;
  • customer data, employee data, internal meeting notes, production identifiers, or private system names;
  • generated exports of a real vault unless manually redacted and reviewed.

Dry-Run-First Rule

Automation should report before it writes. The reference CLI follows this boundary:

  • second-brain capture prints an event by default and writes only with --write.
  • second-brain consolidate --dry-run reports candidates and does not delete, archive, merge, or rewrite notes.
  • destructive cleanup is intentionally outside the public reference CLI.

Reporting a Vulnerability

If you find a security issue, open a private channel with the maintainer before posting details publicly. Include:

  • affected file or command;
  • reproduction steps;
  • why the behavior could expose private data or perform an unsafe action;
  • suggested mitigation if known.

Do not include real secrets, private notes, tokens, or customer data in the report.

Maintainer Checklist Before Publishing

Before pushing changes to this public repository:

  1. Run a sensitive-term scan for private company names, production IDs, credentials, and personal data.
  2. Run python3 -m codex_second_brain.cli validate demo-vault.
  3. Run python3 -m unittest discover -s tests.
  4. Confirm the changed files are under this public package, not the private vault root.

There aren't any published security advisories