Building Agents

Analyze images and frame sequences with OpenAI GPT vision models

Build LLM apps and agents with LangChain memory, tools and workflow patterns

Design conversational assistants with context management and LLM integration

Build production LangChain and LangGraph agents with tools and memory

Analyze images and frame sequences with OpenAI GPT vision models

Load PDDL domains, generate plans with classical planners and validate them

Build research agents with the ToolUniverse Python SDK and its calling patterns

Get help with DeepSeek API calls, model choice and pricing for coding

Check reports and decks against evidence and route risky claims to human review

Build a browser tool to annotate LLM traces and collect structured labels

Categorize LLM pipeline failure modes by reading traces after changes or incidents

Audit an LLM eval pipeline for missing error analysis and unvalidated judges

Measure RAG retrieval and answer quality with synthetic QA and faithfulness checks

Create diverse synthetic test inputs for evaluating an LLM pipeline

Calibrate an LLM judge against human labels using TPR, TNR and bias correction

Design LLM judge prompts for subjective checks like tone, faithfulness and completeness

Fetch current LangGraph documentation for accurate guidance on agent workflows

Provision Entra agent identities and set up OAuth token exchange for agents

Build Foundry IQ knowledge bases that make documents searchable for agents

Build, deploy and evaluate Microsoft Foundry agents, models and hosted agents

Design evaluation datasets and metrics for conversational agents and assistants

Review a dataset for failure modes with a review UI and diverse sample selection

Find the right evals starting point and route to the matching evaluation skill

Write code evaluators that check objective failure rules against agent traces

Turn a repeated code block into a reusable jinx tool registered with the team

Run a saved jinx tool by name or file path with JSON inputs

Benchmark a third-party LLM endpoint for availability, speed, fidelity and quality

Configure Promptfoo evals with LLM-as-judge rubrics and custom Python assertions

Analyze LLM spend by call site and model, then cut costs with smarter profiles

Connect an LLM provider, discover model IDs and verify a model profile from the CLI

Build evals that measure AI product quality with human, code and LLM judges

Run PinchBench to score agent models on calendar, email, coding and research tasks

Audit evals, benchmarks and experiments for leakage and self-confirming results

Pick the cheapest model tier and reasoning effort that can handle a task

Run LLM evaluations on 60+ academic benchmarks such as MMLU, HumanEval and GSM8K

Build Deep Agents with create_deep_agent, harness setup and SKILL.md formatting

Scaffold a minimal local Deep Agent in Python from the official quickstart

Scaffold a minimal local Deep Agent in TypeScript from the official quickstart

Choose between LangChain, LangGraph and Deep Agents before writing agent code

Design agent evals with task specs, Harbor tasks, verifiers and benchmarks

Install LangChain, LangGraph and LangSmith with correct versions and environment setup

Create LangChain agents with tools, middleware and human-in-the-loop handling

Add human approval, custom middleware and structured output to LangChain agents

Scaffold a minimal local LangChain agent in Python from the official quickstart

Build RAG pipelines with document loaders, text splitting, embeddings and vector stores

Scaffold a minimal local LangChain agent in TypeScript from the official quickstart

Scaffold, run, build and deploy LangGraph apps with the langgraph CLI

Swap LLM routing calls for decision models in LangGraph and audit existing graphs

Write LangGraph code with StateGraph, nodes, edges, Command and streaming