AppShield
An agent for the exposed edges of your applications and APIs.
ROADMAP / WHAT WE’RE BUILDINGSix agents.
Six agents.
A deliberate roadmap.
We’re developing a family of focused security products. This is the direction we’re working toward, with room for what we learn from teams along the way.
THE INVESTIGATION MODELCONCEPT / 3D
CONTEXT INEVIDENCE OUTPURPOSE-BUILT SECURITY AGENT.DRAG TO INSPECT
01 / THE PRODUCT DIRECTIONA focused role.
A focused role.
For every agent.
Every product here is in development. Dates, platform support, and release order are still being defined.
In development
Code Guardian
An agent that investigates risk in the context of your codebase.
Mobile Guardian
An agent for the trust boundaries inside your mobile app.
Dependency Armor
An agent that puts dependency risk in the context of your software.
Threat Radar
An agent that connects emerging threats to your environment.
Compliance Vault
An agent for the space between security work and audit evidence.
Real problems
set the order.
A useful first task.
Each agent needs a focused task that produces a reviewable result. We want a narrow workflow to work well before widening its scope.
The right context.
We’re learning which inputs and integrations teams can realistically provide, and where an agent would save investigation time.
Feedback from the work.
Early conversations help shape what we build. Release sequencing will follow what we learn from scoped workflows and early product feedback.
Useful is something
we need to demonstrate.
Know what success means.
Define the task, required inputs, and the decision a reviewer needs to make. Broad promises are hard to evaluate.
Make the result testable.
Connect the recommendation to evidence and make missing context visible. A reviewer should understand what remains uncertain.
Fit into the work.
Test whether the output helps a team take the next step with its existing tools and responsibilities.
These criteria guide development; they are not a release schedule or a claim that the products have passed an evaluation.