Introducing Dogwood: runtime verification for AI agents

I'm Matt Wood, and this is For Your Information. A live list of riffs and links for you and your agent, drawn from what I'm reading, noticing, questioning, concluding, and revising.

Connections

Related to

Agent Session-Aware Runtime Both are runtime infrastructure for AI agents — session-aware runtime handles state continuity while Dogwood handles behavioral correctness verification during that same execution
Strands AI Functions: Python Library for Verified Agent Workflows Both address verification of agent workflows — Dogwood at the runtime monitoring layer, Strands at the function/post-condition enforcement layer; complementary approaches to agent correctness

Supports

Building an Advanced Agentic Harness Dogwood provides a verification layer that would strengthen agentic harness architectures by adding safety guarantees to the orchestration and execution pipeline
Bridging Intent and Execution in Agentic Systems Dogwood directly addresses the intent-execution gap in agentic systems by monitoring whether agent execution actually matches specified safe behaviors at runtime
Strands AI Functions: Post-Conditions and Multi-Agent Teams as Python Functions Dogwood's runtime verification of agent behavior directly supports the post-condition verification model described in Strands, where agent actions must satisfy declared contracts

Develops into

Formal Methods at Jane Street: Agentic Coding Changes the Calculus Dogwood applies formal verification principles (as practiced at Jane Street) to the agentic coding context, operationalizing static formal methods into dynamic runtime checks

Why AI Agents' Deceptive Behavior Concerns Users Dogwood's runtime verification for AI agents directly addresses the deceptive/unreliable behaviors described in the new item — verification is a technical countermeasure to the trust-eroding behaviors of lying, cheating, and unauthorized actions
Cedar Policy Both address runtime control of AI agents - Dogwood via runtime verification, Cedar via policy evaluation - complementary approaches to safe agent execution

Type:	link
First published:	Wednesday, 12 August 2026 at 20:43 UTC
Location:	https://aws.amazon.com/blogs/opensource/introducing-dogwood-runtime-verification-for-ai-agents/
URL:	https://mattwood.fyi/i/0cfd38fe