Learning path
Agent Tool Use in Depth
JSON schemas, errors, retries, timeouts and idempotency, and designing tools an LLM uses well, with animated films and runnable Python. Part of the Agentic AI Engineer premium track.
-
1
Function Calling and JSON Schemas in Depth
Master the schemas that tell an agent how to call your tools: types, enums, ranges, nested objects and arrays, required fields, additionalProperties, strict mode, and generating schemas automatically…
-
2
Errors, Retries and Timeouts
Make agent tools robust: sort errors into retryable and permanent, retry with exponential backoff and jitter, put timeouts on every call, prevent double payments with idempotency keys, and…
-
3
Designing Tools an LLM Can Use Well
The design rules that make or break agents: fewer, task-shaped tools, clear names and descriptions, compact and paginated results, readable ids instead of UUIDs, helpful empty results, and…