X vs Y: every AI-agent tool comparison

Microsoft's framework for multi-agent conversation, with a programming model for agents that talk to each other and tools.

Role-playing autonomous agents that collaborate as a 'crew' with defined roles, goals, and task delegation.

Data framework for connecting LLMs to private data — indexing, retrieval, and agentic RAG over your documents.

Batteries-included agent runtime — built-in memory, knowledge/RAG, and AgentOS, a control plane you run in your own cloud. Formerly Phidata.

Programming — not prompting — language models: compile declarative pipelines into optimized prompts/weights.

Google's Agent Development Kit — a code-first, model-agnostic toolkit for building, evaluating, and deploying multi-agent systems. Optimized for Gemini.

Type-safe agent framework from the Pydantic team — structured outputs, dependency injection, and model-agnostic agents.

OpenAI's lightweight agent framework — a small set of primitives (Agents, Handoffs, Guardrails, Sessions); provider-agnostic via LiteLLM. Evolved from Swarm.

Anthropic's SDK for building agents on Claude — the harness behind Claude Code, with in-process tools, hooks, and an interactive client. Formerly the Claude Code SDK.

AWS's model-driven agent SDK — give it a prompt and tools and let the model plan, call tools, and reflect in a loop. First-class Bedrock, MCP-native, OpenTelemetry tracing.

TypeScript-first SDK for stateful agents where each agent is a Durable Object — embedded SQLite, WebSockets, cron scheduling, and hibernation, running globally on Cloudflare's edge.

A memory layer for AI agents — extracts, stores, and retrieves user/agent facts across sessions.

Long-term memory store for agents with a temporal knowledge graph of facts and their validity over time.

Durable execution platform — write long-running, failure-resilient agent workflows as ordinary code.

Secure cloud sandboxes for running AI-generated code — the runtime layer for code-executing agents.

Test-driven prompt and agent development — evals, red-teaming, and side-by-side model comparison from the CLI.

Pytest-like framework for unit-testing LLM outputs with metrics for hallucination, relevancy, and bias.

The reference collection of Model Context Protocol servers — connect agents to files, GitHub, databases, and more.

Open-source LLM engineering platform — tracing, evals, prompt management, and metrics for agent apps.

Open-source observability for LLM apps via a proxy — logging, caching, and cost tracking with one header.

Cloud-native vector database built for billion-scale similarity search.

High-performance vector search engine with rich filtering, written in Rust for production-scale retrieval.

Vector similarity search inside Postgres — keep embeddings next to your relational data.

In-process analytical database whose vss extension adds an HNSW vector index — vector search alongside your columnar analytics.

Open-source vector database with hybrid search and built-in modules for vectorization and RAG.

Embedded, in-process vector database on the columnar Lance format — versioned, updatable, larger-than-RAM retrieval with no server.