LangChain
Build AI agents with model, tool, and data integrations
An open record format for attributing AI-agent usage and costs

AUDR, short for Agent Usage Detail Record, is an open standard and SDK toolkit for joining usage information across an AI-agent run. A shared run identifier connects business context, model usage, tool activity, and compute measurements from the different systems involved. Defined ownership rules specify which component supplies each field and how records are merged.
Developers can use adapters for supported runtimes or emit records directly through the core SDK, then write them to a local file or another configured sink. The packages are published under Apache 2.0. AUDR records identifiers, usage, and timings rather than prompt content; available adapters and destinations vary, with some integrations still marked as coming soon.
Tried it? Share a useful thought or ask a question.
Good conversations start with a hello.
Sign in to comment