Tool Calling
Letting an LLM call typed functions and APIs in a loop — the mechanism that turns a chat model into an agent.
Tool calling (function calling) lets an LLM emit a structured request to run a function with typed arguments; the runtime executes it and feeds the result back, looping until the task is done. Schemas, validation and stop conditions decide how robust that loop is.
How I use it
- Akashi — my streaming-first tool loop with typed tools and dependency injection, parallel tool calls, composable stop conditions, optional codegen for tool input schemas, and structured output with self-repair.
- mcp-deep-dive — tools exposed and called over the MCP (Model Context Protocol) protocol, with every JSON-RPC frame logged.
- langgraph-ts-agents — Zod-validated tools and structured-output routing.
- Stores Helper — store management operations exposed as tools for coding agents.
Credentials and courses
Developing MCP-Powered Agentic AI Systems (tool use and integrations) and Applied Agentic AI Pipelines with LangChain (ReAct).