All events

Warp event

Trust coding agents with formal verification

Date and time

Thu, Dec 3, 2026 · 2:00 PM EST–3:00 PM EST

As you trust coding agents to write more code unsupervised, it becomes easy for your codebase to run away from you. And patching your skills and prompts with your preferences isn’t going to help.

In this session, we build a verification system that makes undesired code impossible for agents to write. We start with codebase architecture and static checks: structuring your code so bad patterns can’t spread, enforcing team preferences beyond the linter, and running end-to-end verification with computer use before a PR goes up.

Then, we step up from individual checks to rules that always hold. We cover property-based testing to verify behavior across any input, and show how to use formal verification tools like TLA+ to model your application as a state machine and check every possible path.

We provide tools and skills you can bring to any agent workflow, and show how to automate the verification loop using Warp Factories.

Who it’s for: Engineers and engineering leaders running coding agents who want code-level guardrails and formal checks so they can trust agent output at scale.

What you'll learn

  • How to build codebase guardrails that prevent agents from introducing unwanted patterns
  • Which static checks, property-based tests, and end-to-end verification loops make agent output reliable
  • How to model critical application behavior as a state machine and verify every possible path with formal methods
  • How to automate continuous verification for agent-generated code with Warp Factories