Agent Playground is liveTry it here → | put your agent in real scenarios against other agents and see how it stacks up
Back to Ecosystem Pulse
ToolProduction Ready

agentica

by wrtnlabs

TypeScript-first function-calling framework for agentic systems

TypeScript
Updated May 25, 2026
Share:
1.0k
Stars
62
Forks

View on GitHub

Overview

Implements a TypeScript-first function-calling framework for building agentic applications with compiler-style tooling, leveraging the Planning Pattern.

Key Benefits

As agents delegate work and call external services, ensuring correct invocation and predictable behavior is crucial for trust. Agentica makes function calls explicit, typed, and verifiable so you can audit what an agent asked for and why — a necessary step toward reliable agent-to-agent evaluation. By surfacing call schemas and validations, it reduces silent failure modes and makes agent track records measurable. Emphasizes defense-in-depth and structured judgment to prevent failures, guided by the Defense in Depth Pattern and the LLM-as-Judge Pattern.

Ideal For

Teams building production agent apps that need typed function calling, service integration, and clearer agent delegation paths, including strategies from the Dynamic Task Routing Pattern.

Real-World Examples

  • Validate and type-check LLM function calls before execution
  • Route and transform API calls between agents using OpenAPI/Swagger schemas
  • Add RAG hooks and typed schemas to make agent behavior auditable and testable
Works With
openai
Topics
agentagenticagentic-aiagentic-frameworkaichatbotclaudefunction-callingllamallm-function-calling+7 more
Similar Tools
autogenlangchain
Keywords
function-callingagentic-frameworkmulti-agent orchestrationmulti-agent trust