Building Agents

Define isomorphic LLM tools with Zod schemas and approval flows

Build chat, tools, agents, memory and media features in TanStack AI apps

Migrate from the Vercel AI SDK or raw provider SDKs to TanStack AI

Build, debug and review applications on the Cloudflare Agents SDK

Build Foundry agents with hosted tools and MCP servers in Python

Create and manage Azure AI agents with threads, runs and tools in .NET

Create and run Azure AI agents with threads, messages and tools in Java

Call Azure OpenAI chat, embeddings and image models from a .NET app

Manage Azure AI Foundry projects, agents, connections and evaluations in .NET

Manage Azure AI Foundry projects, connections and evaluations from Java

Create versioned Foundry agents, run evaluations and manage connections in Python

Build Foundry project clients for agents, evaluations and connections in TypeScript

Build apps on the GitHub Copilot SDK with custom tools, hooks and MCP servers

Create, localize and deploy declarative agents for M365 Copilot and Teams

Create Entra agent identity blueprints and OAuth2-capable identities for agents

Create and run evaluation suites for M365 Copilot declarative agents

Build multichannel Teams and Copilot Studio agents with ASP.NET Core and MSAL

Build Teams and Copilot Studio agents with streaming responses in Python

Build Teams and Copilot Studio agents with Express hosting and streaming in TypeScript

Choose models by provider health, falling back on rate limits and timeouts

Create agent.yaml configs, SOUL.md personality files and agent folder layouts

Convert agent definitions between OpenAI, CrewAI, Claude Code and other frameworks

Install gitagent and scaffold, configure and validate your first agent

Run semantic search over ingested documents with LlamaIndex or ChromaDB RAG

Configure and run agents through different LLM adapters and switch providers

Build AI features that turn natural language into typed judgments with TypeSafe

Design, audit and troubleshoot provider-neutral agent loops, tool design and harnesses

Design AI automations and agent workflows, picking which manual work to automate

Build apps with the Anthropic SDK and Claude API for Python or TypeScript

Register local or remote model services as managed endpoints for agent runs

Call a registered model endpoint over its native HTTP API for predictions

Analyze audio, images and video and generate media with the Google Gemini API

Find where and why tau2-bench evaluation tasks failed, with root causes per round

Run inference on Fireworks AI directly, via a gateway, LiteLLM or the OpenAI SDK

Log LLM runs with per-turn transcripts, token costs and run comparison reports

Place task-driven phone calls and follow each one through to its result

Look up and verify Claude Agent SDK Python docs and options from the terminal

Call Groq models for chat and transcription with a local token cost ledger

Chat with hosted models through Mixlayer with local PII redaction and cost proofs

Pick the right hosted Ollama cloud model for a prompt and call it

Track OpenRouter spend with budgets, anomaly alarms and failover maps offline

Manage Rightbrain tasks, agents, approvals and evals with a local mirror

Search community AI workflows offline and rank them by weekly ratings

Call xAI Grok models from the terminal through the REST API

Record a Langroid multi-agent design pattern as a note for future reference

Look up design patterns for the Langroid multi-agent framework and its tools

Cut LLM API costs with task-based model routing, budgets, retries and caching

Set up an eval harness to score Claude Code sessions with eval-driven development

Choose between regex and an LLM to parse structured text, LLM only for edge cases