Connectors

Connectors are lightweight relay agents that let Hashiro's AI assessments reach targets inside private networks without requiring VPN access or firewall changes.

How Connectors Work

A connector is a small binary (turing-connector) deployed inside the customer's network. It establishes an outbound connection to the Hashiro control plane, creating a secure relay for assessment agents to reach internal targets.

[AI Agent] → [Hashiro Control Plane] ← [Connector] → [Internal Targets]

Because the connection is outbound, no inbound firewall rules are needed.

Security Model

Feature	Description
Token authentication	Each connector authenticates with a unique token
TOFU key pinning	Trust-on-first-use server key pinning prevents MITM attacks
Token rotation	Tokens can be rotated without redeploying the connector
Self-destruct	Connectors can be remotely deactivated, terminating the relay
Outbound-only	No inbound ports need to be opened

Capabilities

Once connected, the relay supports:

• Command execution: run assessment commands on the internal network
• HTTP proxy: route HTTP requests through the connector to reach internal web applications

Deployment

1. Download the turing-connector binary for your platform
2. Configure it with your organization's connector token
3. Run it on a host inside the target network
4. The connector appears in the Hashiro UI under the project's connector panel

Screenshots use fictional demonstration data.

Monitoring

The Connector Activity panel shows:

• Connection status (online/offline)
• Last seen timestamp
• Activity log of relayed operations

WARNING

Connectors inherit the network permissions of the host they run on. Deploy on hosts with appropriate access to the assessment targets.

Use Cases

• Testing internal web applications not exposed to the internet
• Assessing staging environments behind corporate firewalls
• Evaluating internal APIs and microservices
• Running authenticated assessments against internal identity providers