Agents bundle a model, system instructions, and optional tools so you can industrialise your AI assistants. This section covers instantiation, tool management, structured outputs, streaming, and the shared runtime.Documentation Index
Fetch the complete documentation index at: https://ai.aidalinfo.fr/llms.txt
Use this file to discover all available pages before exploring further.
Key guides
Build an agent– install the package and craft your first generation.Tools & tool loop– declare tools and control the orchestration loop.Structured outputs– validate responses withstructuredOutputand theOutputhelpers.Real-time streaming– consume tokens on the fly and supervise live generations.Shared runtime– store binary resources and expose them to your tools.
Best practices
- Override
systemper call to adapt instructions without reinstantiating the agent. - Tune
maxOutputTokens,temperature,topP, and more directly ingenerateorstream. - Compose specialised agents (drafting, validation, summarisation) to cover an entire flow.
- Combine agents with workflows to orchestrate richer decision loops.