Agent frameworks (no first-party memory layer)

Claude 3.5 Sonnet's beta capability for controlling a computer (mouse, keyboard, screen). Frame for browser-agent and OS-agent products built on Claude.

Originally Microsoft Research's multi-agent framework — fork branded AG2 (2024-10) led by community. Memory is one optional component (AutoGen Memory cross-listed).

Open-source autonomous coding agent for VS Code — terminal + browser + edit operations; uses Anthropic Computer Use style flow.

Open-source AI coding assistant — VS Code / JetBrains; configure-your-own-models. Existing Continue-rules row covers file-backed memory; this row covers the framework.

Multi-agent framework — Crews (sequential) + Flows (event-driven). Open-source core + CrewAI Enterprise. Memory available via Mem0 / Letta adapters (CrewAI Memory cross-listed).

deepset's open-source framework for LLM/RAG/agent apps — Pipelines abstraction; deepset Cloud commercial. Memory is a downstream connector.

Event-driven durable workflow platform — competes with Temporal for the JS/TS/Python crowd. Inngest AgentKit (memory cross-listed) is the agent extension.

Dominant Python/JS framework for LLM apps — tool calling, agents, retrievers, runnables. Memory is one component (LangMem cross-listed); the framework itself is the product.

LangChain's stateful graph runtime for agents — explicit state, persistence, time-travel debugging, human-in-the-loop. Memory is one persistence target (LangGraph Persistence cross-listed).

Data framework for LLM apps — connectors, indices, query engines, agents, workflows. Memory is one of many subsystems (LlamaIndex Memory cross-listed).

Anthropic-led collection of reference Model Context Protocol servers (filesystem, GitHub, Slack, etc.) — substrate beneath the entire MCP-memory ecosystem (Cipher, Memobase, etc.).

Microsoft's enterprise-focused agent framework — multi-language (C#, Python, Java); Plugins / Planners / Memory; integrates with Azure AI / Foundry.

Anthropic-introduced open protocol (Nov 2024) for connecting LLMs to data sources and tools. Now standard across Anthropic, OpenAI, Microsoft, Cursor, Continue, IDEs, etc.

Open-source workflow automation platform with first-class AI Agent / LangChain / Memory / Tool nodes — fastest-growing workflow tool in the AI era.

OpenAI's official Python framework for agents — handoffs, guardrails, tracing built in. Successor to OpenAI Swarm (experimental); officially supported.

Durable-execution platform — Workflows-as-code with crash safety. Used as agent-orchestration backbone at Stripe, Snap, Coinbase, OpenAI Stargate.

Vercel's TypeScript LLM SDK — streaming, tool calls, structured outputs, agents. Dominant framework for Next.js LLM apps.

Google-led open protocol for agent-to-agent collaboration (April 2025) — agent cards, RPCs, server-sent events. Partner companies: Atlassian, MongoDB, Salesforce, ServiceNow, etc.

The first viral autonomous agent demo (March 2023) — GPT-4 + memory + tool use + self-direction. Significant Gravitas raised $12M to commercialise the AutoGPT Platform.

Yohei Nakajima's iconic spring-2023 task-list agent loop — Pinecone-backed memory, OpenAI for execution; ~150 LOC. Influence-disproportionate-to-code.

Authentication and tool-execution platform for AI agents — 100+ apps with OAuth handled, function-call schemas, observability.

ai16z's TypeScript agent framework popularised in the crypto/Web3 agent space — multi-platform (Discord, X, Telegram). Brand evolved into elizaOS.

TypeScript-first agent framework — Workflows, Agents, RAG, eval. Built by the Gatsby team. YC W25.

Multi-agent framework simulating a software company (PM, Architect, Engineer, QA). Famous for its software-development-via-multi-agent demo.

Open-source SWE agent platform (renamed from OpenDevin to OpenHands). All Hands AI raised $5M; widely used as the OSS counterpart to Devin.

Pydantic team's agent framework — type-safe agent definitions, structured outputs, tool calls validated by Pydantic schemas.

Hugging Face's minimal agent library — emphasises code-execution agents (agent writes Python, runs it) over JSON tool-calling. Library is intentionally ~1k LOC.

Open-source JS/TS background-jobs platform — long-running, retries, observability. Trigger.dev v3 (2024) added 'tasks' abstraction for agent workflows.

Agent framework built on OpenAI Assistants API — agencies (roles + comm flow); replaced after Assistants v2 changes.

Cisco + LangChain-led 'Internet of Agents' collective — open standards for discovery, messaging, observability of agent networks.

Authentication and tool platform for AI agents — fully-managed OAuth flows for LLMs to act on user accounts (Gmail, Slack, etc.).

DAGWorks's state-machine framework for agents — state + actions + transitions; lightweight alternative to LangGraph.

One of the original 'agent writes a whole codebase from a prompt' demos — community-maintained; spawned the commercial Lovable (gptengineer.app).

Microsoft Research's generalist multi-agent system — Orchestrator + WebSurfer/FileSurfer/Coder/ComputerTerminal specialists. Built on AutoGen v0.4.

Prefect's LLM toolkit — turns Python functions into LLM-powered ones; structured outputs, classification, extraction. Replaced by Control Flow for agentic workflows in 2024.

Multi-modal agent framework (renamed from Phidata to Agno in 2025) — agents, memory, knowledge, tools; emphasis on multi-modal pipelines.

Durable-execution platform from ex-Confluent / Flink team — Restate Cloud + OSS server; emphasises low-latency invocation vs Temporal's polling.

AWS-released open-source agent framework — released March 2025; model-driven, simple Agent abstraction. Predecessor of AWS Agent SDK / Bedrock AgentCore.

OpenAI Swarm — experimental educational multi-agent framework released Oct 2024; deprecated in favour of the OpenAI Agents SDK in Mar 2025.