Please do not open a public issue for security problems.
Report privately through GitHub: go to the repository's Security tab and choose Report a vulnerability. Include a description, the affected component and version, and steps to reproduce. If you have a proof of concept, attach it to the report rather than publishing it.
We aim to acknowledge reports within 3 working days and to agree on a disclosure timeline with you. We will credit you in the release notes unless you prefer otherwise.
The Hub is often reachable from the internet, so we are especially interested in:
- Hub authentication — bypassing the web password, passkeys or session cookies; using a revoked device token; brute-forcing past the login rate limit.
- Injection into agents — anything that lets someone other than the owner get text injected into an agent's context through hooks or MCP, or route a decision to the wrong session.
- The device side — anything that lets another local user or process talk to the daemon's socket, read device tokens, or make
ks hookblock or alter an agent. - Uploads and the web UI — path traversal, stored XSS through requirements, comments, evidence or file names.
- Install script and releases — anything that lets a tampered build pass the SHA-256 check.
Out of scope: denial of service, issues that require an already compromised machine or Hub account, and missing hardening headers without a demonstrated impact.
Kitestring is pre-1.0. Security fixes are released for the latest version only.