Skip to content

Security: ailelabs/aile.sh

SECURITY.md

Security

This client runs unattended on machines we do not control, holds an account token, and updates itself. Treat a flaw in it as a flaw that reaches every lender at once.

Reporting a vulnerability

Use GitHub's private reporting: Report a vulnerability. It opens a private thread with the maintainers and stays closed until there is a fix to publish.

Email security@aile.sh if you cannot use GitHub.

Please do not open a public issue, a pull request, or a discussion for a vulnerability. A public report starts a race between the fix and whoever reads it first.

What to expect: an acknowledgement within 3 working days, an assessment within 10, and credit in the advisory and changelog unless you ask us to leave your name out.

What to include

A report we can reproduce gets fixed faster than one we have to guess at:

  • The version (aile --version), your OS, and your Node version.
  • What an attacker gains, and what they need to start.
  • Steps to reproduce, or a proof of concept.
  • Any log output, with your token and any API key removed.

In scope

  • Anything that makes the node dial a host outside the egress allowlist, or reach a private address (127.0.0.1, 192.168.x, 169.254.169.254).
  • Anything that writes a provider credential to disk, prints one, or sends one somewhere other than the configured relay.
  • Anything that lets a party other than you read or change your config, account token, or node identity.
  • A path that lets a malicious relay response run code on the node.
  • A flaw in the update mechanism that installs something other than the published aile.sh package.
  • Anything that lets one lender read another's traffic, earnings, or accounts.

Out of scope

  • Reading prompts on a self-hosted model. aile local runs a model on your machine, so your machine reads those prompts. That is how the feature works and the client says so at every point it applies.
  • What the aile.sh server can see. The server terminates TLS with the provider, so it reads prompts and responses. See the privacy model. Report server-side findings against the server, not here.
  • Vulnerabilities in Node, npm, or a provider's own API.
  • Attacks that need root on the machine already, or physical access to it.
  • A denial of service against your own node.
  • Missing hardening with no path to exploit behind it.

Scope of this file

SECURITY.md covers the aile.sh npm package and this repository. For the relay, the web app, or the API, write to security@aile.sh instead.

There aren't any published security advisories