Thank you for taking the time to report a security problem responsibly. It is genuinely appreciated.
| Version | Security fixes |
|---|---|
| Latest release (1.x) | ✅ Yes |
| Previous minor version | ✅ Critical only |
| Anything older | ❌ No — please update |
Do not open a public issue, discussion or pull request for a security problem. A public report exposes every installation until a fix ships.
Use GitHub's private reporting:
The report is private between you and the maintainer, needs no email exchange, and becomes a published advisory with credit to you once a fix ships. A GitHub account is required; creating one is free.
If you cannot use GitHub at all, open a Discussion saying only that you have a security report — no details — and you will be given a private channel.
- The version of BasicInventory and your operating system.
- What the vulnerability allows an attacker to do, and what access they need first (local user, network, a file the victim opens…).
- Step-by-step reproduction, ideally with a minimal proof of concept.
- Any log excerpts or screenshots — with your business data removed.
| When | What |
|---|---|
| Within 48 hours | Acknowledgement that the report arrived. |
| Within 7 days | An initial assessment: confirmed or not, and a severity. |
| Ongoing | Progress updates at least every two weeks. |
| On release | A fixed version, an advisory, and credit to you unless you prefer to stay anonymous. |
Target fix windows, from confirmation: critical 14 days, high 30 days, medium or low with the next scheduled release.
In scope — the BasicInventory desktop application distributed from this repository's Releases:
- Remote or local code execution, privilege escalation.
- Extraction of the encrypted AI provider key, or of inventory data, by another user or process on the same machine.
- Tampering with the update mechanism (serving a forged update).
- Injection into the AI assistant's read-only data layer that reaches tables or files it must not reach.
- Anything that sends user data off the machine other than the two documented cases (update check, and the AI provider you configured).
Out of scope:
- Vulnerabilities that require an attacker to already be an administrator on the machine, or physical access to an unlocked session.
- The unsigned installer triggering a SmartScreen warning — known and planned.
- The AI provider's own service, its models or its billing.
- Reports produced only by an automated scanner, with no demonstrated impact.
- Denial of service against your own local installation.
- Social engineering of the maintainer or of users.
Research done in good faith under this policy will not be met with legal action. Please: test only against your own installation, do not access, modify or exfiltrate anyone else's data, do not degrade a third party's service, and give a reasonable window to fix before disclosing.
Some things people report as vulnerabilities are deliberate properties, and are documented so you can check them:
- Your inventory never leaves the machine. There is no telemetry, no analytics and no crash upload.
- The AI assistant is off until you configure it, is strictly read-only, and its data access is an allowlist of inventory tables — configuration, keys and the file system are unreachable.
- Your AI provider key is encrypted (AES-256-GCM) with a machine-local key kept outside the database, so a copy of the database file carries no usable credential.
If you can demonstrate that any of those statements is false, that is exactly the kind of report this policy exists for.