Introducing two features to the OpenRouter Agent SDK: lifecycle hooks and async tools!
If you aren't familiar, the Agent SDK is our toolkit for building agentic applications with OpenRouter. These two additions allow to build even more robust AI workflows.
Lifecycle hooks let you take specified actions at fixed points in the callModel loop.
Block destructive commands, track token usage and cost, redact secrets from prompts, or mutate tool arguments before execution.
Here's a simple example using PreToolUse to block a dangerous command.
Full hook list and more info on how to use them in the docs: https://t.co/f4Ri4NqxGi
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
agentic loop — The cycle an agent runs in: decide, call a tool, read the result, decide again — repeating until the goal is met or a stop condition fires.
Why it matters
Hooks provide an enforcement point for policy inside the agentic loopThe cycle an agent runs in: decide, call a tool, read the result, decide again — repeating until the goal is met or a stop condition fires.Full definition → rather than around it, and deferred tools let a run pause for human review and resume from a webhook hours later in another process.
