A small, verified index of agentic AI tooling for builders

An open-source framework and runtime from LangChain for building stateful agents as graphs that mix deterministic code with LLM-driven steps. Provides durable execution, persistence, human-in-the-loop interrupts, streaming and memory.

2026-10-22

CrewAI

Microsoft Agent Framework

An open-source Python framework for orchestrating teams of agents that each hold a role, goal, backstory and tools, executing through sequential or hierarchical Crews and event-driven Flows. A separate commercial platform (CrewAI AMP) adds a visual editor and governance.

2026-10-22

Microsoft's unified agent SDK and the designated successor to both AutoGen and Semantic Kernel, reaching 1.0 GA on 3 April 2026. Combines AutoGen's multi-agent abstractions with Semantic Kernel's enterprise connectors and observability, adding typed graph-based workflows.

2026-10-22

OpenAI Agents SDK

OpenAI's lightweight open-source framework built from a small set of primitives: Agents, Handoffs, Guardrails, plus Sessions memory and built-in tracing. A production-ready successor to the earlier Swarm experiment.

2026-10-22

LlamaIndex

An open-source Python framework for building RAG and agentic applications over private data, with connectors, indexes, query engines, agents and event-driven Workflows. Companion commercial services LlamaParse and LlamaCloud add document parsing and managed retrieval.

2026-10-22

Pydantic AI

A model-agnostic Python agent framework from the Pydantic team, built for type-safe agents with validated structured outputs, function tools, dependency injection and graph-based control flow. Reached 1.0 in September 2025.

2026-10-22

Mastra

An open-source TypeScript framework for agents, durable typed workflows, RAG, memory, MCP, evals and observability, built by the team behind Gatsby. A commercial platform adds hosted deployment, a visual Studio and enterprise controls.

2026-10-22

DSPy

Google Agent Development Kit (ADK)

An open-source Python framework from Stanford NLP for programming rather than prompting language models: you declare typed Signatures and use Modules and Optimizers to compile programs against a metric. Actively developed (v3.4.1, October 2026).

2026-10-22

Google's open-source, code-first framework for building, evaluating and deploying agents, covering single agents through multi-agent and graph-based workflows. Optimised for Gemini but model- and deployment-agnostic.

2026-10-22

GitHub MCP Server

GitHub's official MCP server exposing repository, issue, pull request, file, code-search and workflow actions to MCP clients. Runs locally over stdio, supports a Streamable HTTP mode, and is also hosted remotely by GitHub.

2026-10-22

GitLab MCP Server

GitLab's built-in MCP server connecting external agents to projects, issues, merge requests, pipelines and work items. Generally available since GitLab 19.5 across all editions including Self-Managed and Dedicated.

2026-10-22

Playwright MCP

Playwright's official MCP server giving agents real browser control through structured accessibility snapshots rather than pixels, so no vision model is required. Tools cover navigation, snapshots, clicking, typing and network inspection.

2026-10-22

Browserbase MCP

An MCP server that gives any client a cloud browser, built on Browserbase and Stagehand so agents can act on pages in natural language. Offers a hosted Streamable HTTP endpoint and a self-hostable stdio version.

2026-10-22

Supabase MCP

Cloudflare API MCP Server

Supabase's official hosted MCP server connecting assistants to Supabase projects to manage tables, run SQL, handle migrations and branches, fetch configuration and read logs. Exposes 20+ tools in configurable feature groups.

2026-10-22

Cloudflare's managed remote MCP server providing access to the wider Cloudflare API across DNS, Workers, R2 and Zero Trust, plus developer documentation. Uses a search-and-execute pattern where the model writes code against the API schema, run in an isolated sandbox.

2026-10-22

AWS MCP Servers

AWS's suite of roughly 50 open-source MCP servers, each scoped to one service or workflow such as CDK, Serverless/SAM, EKS, ECS, Bedrock Knowledge Bases and cost analysis. AWS also offers managed successors including a free AWS Knowledge MCP Server for documentation.

2026-10-22

Stripe MCP

Stripe's remote MCP server exposing payment API tools and knowledge-base search so agents can read and write Stripe data and run analytics. Authenticates via OAuth for interactive clients or dedicated agent keys for persistent agents.

2026-10-22

Sentry MCP

Sentry's hosted MCP server connecting assistants to Sentry for error search, performance analysis, issue triage, documentation and project management. All connections use OAuth and endpoints can be scoped to an organisation or project.

2026-10-22

Composio Connect

A hosted MCP aggregator giving any MCP client access to 1,000+ apps through a single connection. Rather than exposing every app tool directly, it presents a small set of meta-tools for discovery, on-demand OAuth authorisation and parallel execution.

2026-10-22

Filesystem MCP

The official reference MCP filesystem server, giving agents read and write access restricted to explicitly allowed directories. Exposes tools for reading, writing, editing, moving, listing and searching files, rejecting paths outside the allowed set.

2026-10-22

LangSmith

LangChain's framework-agnostic observability and evaluation platform for LLM applications and agents, providing run tracing, dashboards, prompt management and automated evals. Offers managed cloud, bring-your-own-cloud and self-hosted options.

2026-10-22

Langfuse

An open-source (MIT) agent evaluation and observability platform covering tracing, prompt management, datasets, experiments and evals, with a free self-hostable core. OpenTelemetry-native with 100+ integrations.

2026-10-22

Braintrust

A hosted observability, evaluation and gateway platform for LLM applications, offering multi-step agent tracing, LLM-as-a-judge and custom scorers, regression testing in CI and production monitoring.

2026-10-22

Arize Phoenix

An open-source AI observability and evaluation platform that traces model calls, retrievals, tool use and custom logic to debug agents. Built on OpenTelemetry and the OpenInference spec, runnable locally, self-hosted, on Kubernetes or via managed cloud.

2026-10-22

promptfoo

An open-source CLI and library for systematic LLM evaluation, prompt testing and AI security, running YAML-defined test cases across 50+ providers with automated scoring, plus red-teaming for prompt injection and data leakage.

2026-10-22

DeepEval

An open-source evaluation framework styled as Pytest for LLM applications, with 50+ research-backed metrics including LLM-as-a-judge, G-Eval and hallucination detection, plus synthetic data generation.

2026-10-22

E2B

An open-source cloud platform that runs one isolated microVM per agent session for executing AI-generated or untrusted code, with snapshots, shells, filesystems, browsing, networking and pause/resume. Supports both its own cloud and bring-your-own-cloud.

2026-10-22

Daytona

A sandboxed code-execution platform for AI-generated code and agent workflows, providing container-based sandboxes created in under 90ms, stateful persistent workspaces, environment snapshots and optional GPU compute. Available as a managed cloud with customer-managed compute.

2026-10-22

Modal

A serverless cloud for AI workloads that runs Python functions, model endpoints, batch jobs and isolated Sandboxes for executing untrusted agent code, with per-second billing and scale-to-zero autoscaling.

2026-10-22

Letta