aidd is a local-first tool: it runs on your machine, against your projects, with the backend providers and credentials you configure. Even so, we take security issues seriously and appreciate responsible disclosure.

For a full inventory of what aidd reads, writes, runs, and sends over the network (including the .aidd/ write allowlist, recipe shell steps, and backend network egress), see What aidd Can Modify.

We make security fixes against the latest released version. Please reproduce issues on the most recent release before reporting.

Please do not open a public issue for a security vulnerability.

Use GitHub's private vulnerability reporting (Security → Report a vulnerability on the repository) to open a confidential advisory. Include:

• a description of the issue and its impact,
• steps to reproduce or a proof of concept,
• affected version(s) and platform,
• any suggested remediation.

We aim to acknowledge a report within a few days and to keep you updated as we investigate and prepare a fix. We will credit reporters in the release notes unless you prefer to remain anonymous.

aidd executes AI agent workflows that can read and modify the projects you point it at, run shell steps in recipes, and (when you opt in) bind the web panel to your LAN. When assessing security, keep in mind:

• The web panel binds to 127.0.0.1 by default. Remote binding is opt-in and should be paired with web.authToken; see docs/guides/deployment.md.
• Provider API keys and channel tokens live in ~/.aidd/config.json and are never exposed through the web settings surface.
• Recipes can run shell steps and agents can modify files; treat the projects and recipes you run as trusted input.

Reports that require already-untrusted local access (for example, an attacker who can already read ~/.aidd/config.json) are generally out of scope, but we still welcome a heads-up.