Please report security issues privately through GitHub Security Advisories rather than a public issue.
Include what you did, what happened, and what you expected. A proof of concept helps a lot. I will acknowledge the report and keep you updated as it is fixed.
This is a hobby project maintained in spare time, so please do not expect an enterprise response time — but security reports go to the front of the queue.
Being honest about the threat model saves everyone time:
- A tunnel publishes a local service to the internet. That is the whole point of the tool. Anyone who knows the URL reaches your service; burrow does not add authentication in front of it. If the thing you tunnel has no auth, neither does the tunnel.
- Agent tokens are bearer credentials. Whoever holds one can publish
tunnels as that token. They cross the network in the clear unless the control
listener has TLS (
-tls-cert/-tls-key) — configure it on any server that is not on a trusted private network. - The admin panel is a single-password gate. It is meant for one operator,
not for a team with roles. If you would rather not expose it at all, leave it
off the public host and bind it to loopback with
-admin-addr 127.0.0.1:7002, reaching it over an SSH port-forward. -insecuredisables certificate verification. The connection stays encrypted but is no longer authenticated, so it is only reasonable while setting up a self-signed certificate.- Tunnel operators can see tunnel traffic. The server proxies plaintext HTTP to the agent; end-to-end secrecy from the server operator is not a property this design has.
Reports that fit the model above are very much wanted, for example:
- Bypassing admin authentication or the CSRF protections on the API.
- One token claiming a subdomain or port reserved by another.
- Reaching a tunnel or the panel from a host that should not route there.
- Crashes or resource exhaustion triggered by a remote, unauthenticated peer.
- Anything that lets an agent read or affect another agent's traffic.
- Missing authentication on whatever you chose to tunnel.
- Attacks that require the admin password or a valid agent token you were given.
- Denial of service by an authenticated agent that you yourself issued a token to (revoke it in the panel).