Canopy FAQs

Question 1

Can I define complex access rules with Canopy?

Accepted Answer

Yes, Canopy supports defining flexible 'flows' with allowed tool calls (including regexes). For advanced scenarios, it offers multi-flow policies, like the 'Rule of Two', to limit the number of concurrently allowed tool interactions, greatly reducing the attack surface.

Question 2

How does Canopy protect against prompt injection and jailbreak attacks?

Accepted Answer

Canopy blocks tool calls that violate defined policies. Additionally, it integrates with PromptGuard2 to probabilistically detect and block tool responses that contain prompt injection attacks or jailbreaks, adding a crucial layer of defense.

Question 3

What is Canopy and how does it enhance agentic workflow safety?

Accepted Answer

Canopy is a tool that defines and enforces interaction policies for agentic clients, significantly improving the safety of complex workflows. It prevents misuse by blocking unauthorized tool calls based on predefined rules.

Question 4

Does Canopy help with secure credential management for agentic tools?

Accepted Answer

Absolutely. Canopy supports secure storage and injection of secrets using your operating system's keyring. This means sensitive API keys and credentials are not stored in plain text, enhancing overall security for your MCP configuration.

Question 5

Is there a way to monitor agentic tool calls for compliance?

Accepted Answer

Yes, Canopy offers an auditing mode. When enabled, it sends webhook notifications for every tool call, providing visibility and a clear audit trail of agentic actions and policy adherence, both in proxy and hook modes.

Canopy

Canopy

Características Principales

Casos de Uso

Características Principales

Casos de Uso