Agents

Agent turns call paid models and may change external systems through tools. Pyrula records completed LLM and tool calls so crash recovery does not execute them again.

pyrula.agents runs agents on the durable workflow engine. An interrupted turn resumes from its journal, and completed calls return their recorded results. The package also provides streaming and HTTP serving.

from pyrula.agents import agent, tool

Start with the Quickstart to define an agent and prove a crash resumes without re-running work. For the execution model agents inherit (runs, replay, suspend points), read the Workflows concepts.

Guides

• Authoring: the @agent and @tool decorators, calling an LLM, and the context surface.
• Integrations: bring LangChain tools into a Pyrula turn, and run a durable Pyrula workflow as a LangGraph node.
• Serving: run agents over HTTP, talk to them with the client, scale execution on workers, drive it from the CLI, and fire them on schedules.
• Observability: OpenTelemetry tracing and metrics with GenAI semantic conventions.