Skip to content

Security: kadevin/ilab-gpt-conjure

Security

SECURITY.md

Security Policy

Local-only assumptions

iLab GPT Conjure is designed for local personal workflows. Do not expose the WebUI directly to the public internet unless you have reviewed and hardened the deployment yourself.

Secrets and local data

Do not publish OAuth tokens, API keys, account files, .env files, input images, generated outputs, task metadata, SQLite databases, or debug logs.

Sensitive local paths include:

  • ~/.codex/auth.json
  • output/
  • outputs/
  • input/
  • inputs/

Advanced local auth warning

The optional Codex / ChatGPT OAuth mode calls an internal ChatGPT backend endpoint. It is not an officially recommended OpenAI API integration path and may change or stop working without notice. Prefer OpenAI-compatible API mode for stable integrations.

Portable updater behavior

Portable startup launchers only start the local WebUI server and open the local browser URL. They do not contact GitHub and do not update files automatically.

Portable update scripts are manually run. They fetch GitHub Release metadata and the matching portable zip, verify the published SHA256 file, preserve local data/, only replace package-managed files inside the extracted portable folder, and keep backups under .backup/.

Reporting issues

Please report security issues privately to the maintainer instead of opening a public issue containing credentials, tokens, private prompts, or private images.

There aren't any published security advisories