The infrastructure AI agents run on.

Cloud browsers, sandboxes, memory, search, email, payments and voice. One checked list for people shipping agents, not demos.

84tools listed

13categories

243side-by-side comparisons

Oct 31last full link check

Covering Browserbase, E2B, Mem0, Exa and 80 more

Browse by layer of the stack

All 84 tools

Browsers & Computer UseCloud browsers and SDKs that let agents use the web.10 tools

Code SandboxesIsolated machines where agents can safely run code.13 tools

Memory & ContextLong-term memory layers so agents remember users and tasks.8 tools

Search & Web DataAPIs that give agents fresh web search and clean page content.9 tools

EmailInboxes and sending APIs for agents that talk over email.6 tools

Integrations & Auth ToolsPrebuilt tool calls and managed OAuth for thousands of apps.6 tools

PaymentsWays for agents to charge, pay, and hold credentials.3 tools

Voice AgentsPlatforms for phone and voice agents.3 tools

Observability & EvalsTracing, debugging, and evals for agents in production.6 tools

Agent FrameworksLibraries for building agents and multi-agent workflows.6 tools

IdentityLogin, permissions, and token handling for agents.1 tools

Durable ExecutionWorkflow engines that let long-running agents resume after a crash.6 tools

Model GatewaysOne API in front of many models, with routing, fallbacks and spend controls.7 tools

Featured

Featured and sponsored

Paid placements, always labeled and kept out of the organic list.

Get featured

Featured listing

Your tool here

A labeled top spot in your category, a featured badge, and a mention in the next Friday email. Live within 24 hours of review.

Get featured for $99/month

Just added

New this week

All new

Desearch
desearch.ai

Search API for AI agents: one API key covers web search, AI search with source links, page extraction to clean text or markdown, and X search. REST API, Python and JavaScript SDKs, and a hosted MCP server. Usage-based pricing.
Search & Web DataNew
Tokenrip
tokenrip.com

Shared workspace your AI agents work from: files with version history, tables, tasks any agent can pick up, and pinned decisions every agent reads when it loads the workspace. Agents connect from Claude, ChatGPT, Codex, Cursor or any MCP app through a hosted MCP server, a CLI or an HTTP API, each with its own revocable key, and every change is logged with the agent that made it.
Memory & ContextNew
Browserless
browserless.io

Cloud browser automation platform: connect Puppeteer, Playwright or Selenium to managed headless browsers, or use BrowserQL, REST APIs and an MCP server for scraping, screenshots, PDFs and AI-driven browsing. Self-hosting available, with a paid license for commercial use.
Browsers & Computer UseNew
Lightpanda
lightpanda.io

Open-source headless browser written from scratch in Zig for automation, crawling and AI agents: runs JavaScript without graphical rendering, works with Puppeteer and Playwright over CDP, with a hosted cloud option.
Browsers & Computer UseOpen sourceNew
MemorySync
memorysync.io

MemorySync is a memory API for AI applications: your agent writes what it learns about each user, and MemorySync returns the relevant parts later, ranked by meaning and scoped to that user.
Memory & ContextNew
MepMail
mepmail.dev

Email for apps and AI agents: a Resend-compatible sending API with SMTP, plus mailboxes on your own domain that agents read and draft in over MCP, where you choose direct sending or approval for each reply. AGPL open source and self-hostable.
EmailOpen sourceNew
MachineMail
machinemail.dev

Email API for AI agents: free inboxes on a shared domain or your own, a trust report on every inbound message, a REST API that follows AgentMail's paths and field names, webhooks and an MCP server.
EmailNew
Runtime
withruntime.com

Linux sandboxes for agent code, each a Firecracker microVM, billed on active CPU, with pause, forks, snapshots, custom images, an MCP server and SDKs.
Code SandboxesNew

Browse the stack

84 tools

Agent 37
agent37.com

Cloud sandboxes for running AI agents, positioned as lower-cost agent infrastructure.
Code Sandboxes
AgentMail
agentmail.to

Email inbox API for agents: create, send, and receive real inboxes.
Email
AgentOps
agentops.ai

Tracing and debugging for AI agents.
Observability & Evals
Aident Loadout
aident.ai

Capability layer connecting agents to 1,000+ apps and 27,000+ actions, with a credential vault, audit log, CLI and MCP server.
Integrations & Auth Tools
AIO Sandbox
sandbox.agent-infra.com

All-in-one agent sandbox with browser, shell, file access, VSCode Server, and MCP in one container.
Code SandboxesOpen source
Anchor Browser
anchorbrowser.io

Infrastructure for computer-use agents that automate web tasks at scale.
Browsers & Computer Use
Apify
apify.com

Marketplace of ready-made scrapers and automation tools callable by agents.
Search & Web Data
Arcade
arcade.dev

Actions runtime that lets agents act on users' behalf with governed auth.
Integrations & Auth Tools
Auth0 for AI Agents
auth0.com

Auth0 tools for user login, token vaults, and permissions in AI agents.
Identity
Braintrust
braintrust.dev

Observability and evals platform for production agents.
Observability & Evals
Brave Search API
brave.com

Independent web search API covering a 40B+ page index.
Search & Web Data
Browser Use
browser-use.com

Open-source browser agent library plus hosted remote browsers for AI agents.
Browsers & Computer UseOpen source
Browserbase
browserbase.com

Hosted headless browsers for agents, with session control and an MCP server.
Browsers & Computer Use
Browserless
browserless.io

Cloud browser automation platform: connect Puppeteer, Playwright or Selenium to managed headless browsers, or use BrowserQL, REST APIs and an MCP server for scraping, screenshots, PDFs and AI-driven browsing. Self-hosting available, with a paid license for commercial use.
Browsers & Computer UseNew
Celesto
celesto.ai

Hardware-isolated cloud computers for AI agents, with browser, desktop, shell and file access, pause, resume and forking, via CLI and SDK; Apache-2.0 microVM core you can self-host.
Code SandboxesOpen source
Cloudflare AI Gateway
developers.cloudflare.com

Analytics, caching, rate limiting and model fallback in front of your AI provider calls.
Model Gateways
Cloudflare Sandbox SDK
developers.cloudflare.com

Run untrusted or generated code in Linux VMs or isolated Workers on Cloudflare.
Code Sandboxes
Cognee
cognee.ai

Open-source agent memory platform built on graph and vector stores.
Memory & ContextOpen source
Coinbase x402
x402.org

Open standard for internet-native payments using HTTP 402, built for agents.
Payments
Composio
composio.dev

Tool calls with delegated auth across 1,500+ apps for agents.
Integrations & Auth Tools
ContextNest
promptowl.ai

Versioned markdown context vault for agents, read through a CLI or an MCP server.
Memory & ContextOpen sourceNew
Cosmonic
cosmonic.com

WebAssembly sandboxes for agent code, MCP servers and tools, with no access until you grant it: the free Cosmonic Desktop app (public beta) and Cosmonic Control for Kubernetes. Open source core: CNCF wasmCloud.
Code Sandboxes
CrewAI
crewai.com

Framework and platform for multi-agent teams.
Agent FrameworksOpen source
Daytona
daytona.io

Secure sandboxes for running AI-generated code. Its public GitHub repo is archived: core development moved to a private codebase in June 2026.
Code Sandboxes
DBOS
dbos.dev

Open-source durable execution and workflow orchestration system.
Durable ExecutionOpen source
Desearch
desearch.ai

Search API for AI agents: one API key covers web search, AI search with source links, page extraction to clean text or markdown, and X search. REST API, Python and JavaScript SDKs, and a hosted MCP server. Usage-based pricing.
Search & Web DataNew
E2B
e2b.dev

Isolated cloud machines for agents to run code, browse, and use tools.
Code SandboxesOpen source
ElevenLabs Agents
elevenlabs.io

Build and deploy chat and voice agents on ElevenLabs.
Voice Agents
Exa
exa.ai

Search API built for AI agents, with real-time web data and deep research.
Search & Web Data
Firecrawl
firecrawl.dev

Turns websites into clean, LLM-ready data with crawl, scrape, and search APIs.
Search & Web DataOpen source
Hatchet
hatchet.run

One platform to orchestrate AI agents, schedule background tasks, and run workflows.
Durable ExecutionOpen source
Helicone
helicone.ai

AI gateway with routing and LLM observability.
Observability & EvalsOpen source
Hyperbrowser
hyperbrowser.ai

Cloud Chrome sessions, isolated sandboxes, and web data APIs for agents.
Browsers & Computer Use
Inngest
inngest.com

Durable execution for background jobs, workflows, and AI agents, with automatic retries, flow control, and step-level observability.
Durable Execution
Jina Reader
jina.ai

Converts any URL to Markdown for grounding LLMs.
Search & Web Data
Kernel
kernel.sh

Browser infrastructure for web agents with reusable sessions and fast spin-up.
Browsers & Computer Use
Kong AI Gateway
konghq.com

Security, routing, observability and cost control for LLM and MCP traffic, from the Kong API gateway team.
Model GatewaysOpen source
Langfuse
langfuse.com

Open-source tracing, evals, and observability for agents.
Observability & EvalsOpen source
LangGraph
langchain.com

Agent orchestration framework with durable execution and human-in-the-loop.
Agent FrameworksOpen source
LangSmith
langchain.com

LangChain's tracing and monitoring platform for agents and LLM apps.
Observability & Evals
Letta
letta.com

Platform for stateful agents that keep long-term memory.
Memory & ContextOpen source
Lightpanda
lightpanda.io

Open-source headless browser written from scratch in Zig for automation, crawling and AI agents: runs JavaScript without graphical rendering, works with Puppeteer and Playwright over CDP, with a hosted cloud option.
Browsers & Computer UseOpen sourceNew
LiteLLM
litellm.ai

AI gateway and LLM proxy: one OpenAI-compatible key for 140+ providers, with spend tracking, caps and routing. Self-hostable.
Model GatewaysOpen source
MachineMail
machinemail.dev

Email API for AI agents: free inboxes on a shared domain or your own, a trust report on every inbound message, a REST API that follows AgentMail's paths and field names, webhooks and an MCP server.
EmailNew
Mailgun
mailgun.com

Email API to send, receive, and track email.
Email
Mastra
mastra.ai

TypeScript framework for agents with memory, tools, MCP, and workflows.
Agent FrameworksOpen source
Mem0
mem0.ai

Memory layer that lets agents learn from past interactions.
Memory & ContextOpen source
MemorySync
memorysync.io

MemorySync is a memory API for AI applications: your agent writes what it learns about each user, and MemorySync returns the relevant parts later, ranked by meaning and scoped to that user.
Memory & ContextNew
MepMail
mepmail.dev

Email for apps and AI agents: a Resend-compatible sending API with SMTP, plus mailboxes on your own domain that agents read and draft in over MCP, where you choose direct sending or approval for each reply. AGPL open source and self-hostable.
EmailOpen sourceNew
Modal
modal.com

Serverless cloud for running, training, and serving AI workloads, including sandboxes.
Code Sandboxes
Nango
nango.dev

Integration platform connecting products and agents to 1,000+ APIs.
Integrations & Auth Tools
OpenAI Agents SDK
openai.github.io

OpenAI's lightweight Python framework for multi-agent workflows.
Agent FrameworksOpen source
OpenRouter
openrouter.ai

One API for many models from many providers, with model and price comparison.
Model Gateways
Parallel
parallel.ai

Web search, extraction, and monitoring APIs built for AI.
Search & Web Data
Pickle Browser
picklebrowser.com

Local browser that opens real pages for agents, shows its work on your machine, and needs no account.
Browsers & Computer Use
Pipedream Connect
pipedream.com

Managed auth and thousands of prebuilt tools across 3,000+ APIs.
Integrations & Auth Tools
Portkey
portkey.ai

AI gateway and production stack for teams shipping generative AI apps, with an open-source gateway.
Model GatewaysOpen source
Pragor
pragor.net

Operations board for teams of AI agents: shared tasks and messages, human approvals for risky steps, an attributable audit trail, app health reporting and an MCP server, across any framework or model.
Observability & Evals
Pydantic AI
pydantic.dev

Type-safe Python agent framework from the Pydantic team.
Agent FrameworksOpen source
Requesty
requesty.ai

OpenAI-compatible LLM router for 600+ models with smart routing, caching, failover and EU data residency.
Model Gateways
Resend
resend.com

Developer email API for transactional and marketing mail.
Email
Restate
restate.dev

Runtime for agents and backends with durable execution, state, queues, RPC, timers, and flow control in one binary.
Durable Execution
Retell AI
retellai.com

Voice agents for inbound and outbound phone calls.
Voice Agents
Runloop
runloop.ai

Code sandboxes (Devboxes) for AI coding agents, plus evaluations.
Code Sandboxes
Runtime
withruntime.com

Linux sandboxes for agent code, each a Firecracker microVM, billed on active CPU, with pause, forks, snapshots, custom images, an MCP server and SDKs.
Code SandboxesNew
Sandbox0
sandbox0.ai

Cloud computers for AI agents that keep the whole writable root filesystem, pause compute between tasks and resume on demand, via API, SDKs and CLI.
Code Sandboxes
Shipmail
shipmail.to

Business email hosting with an API for agents: custom-domain mailboxes agents can send from and read, webhooks, TypeScript and Python SDKs, a CLI and an MCP server.
Email
Skyfire
skyfire.xyz

Identity and payment credentials for AI agents on the open web.
Payments
Smithery
smithery.ai

Registry and hosting to connect agents to MCP tools and services.
Integrations & Auth Tools
Stagehand
stagehand.dev

SDK for browser agents with a Playwright-style API and self-healing selectors.
Browsers & Computer UseOpen source
Steel
steel.dev

Open-source browser API for running fleets of cloud browsers for agents.
Browsers & Computer UseOpen source
Stripe Agent Toolkit
docs.stripe.com

Stripe's MCP server, plugins, and skills for agents that handle payments.
Payments
Supermemory
supermemory.ai

Memory and continual learning engine for agents.
Memory & Context
Tavily
tavily.com

Real-time search and content extraction APIs for agents.
Search & Web Data
Temporal
temporal.io

Open-source durable execution platform where you write workflows as code and they pick up where they left off after a failure.
Durable ExecutionOpen source
Tokenrip
tokenrip.com

Shared workspace your AI agents work from: files with version history, tables, tasks any agent can pick up, and pinned decisions every agent reads when it loads the workspace. Agents connect from Claude, ChatGPT, Codex, Cursor or any MCP app through a hosted MCP server, a CLI or an HTTP API, each with its own revocable key, and every change is logged with the agent that made it.
Memory & ContextNew
Trigger.dev
trigger.dev

Open-source platform for durable AI agents and workflows in TypeScript, with long-running tasks, retries, queues, and LLM observability.
Durable ExecutionOpen source
Vapi
vapi.ai

Developer platform to build, test, and deploy voice AI agents.
Voice Agents
Vercel AI Gateway
vercel.com

Use 400+ AI models through one endpoint without managing rate limits and API keys per provider.
Model Gateways
Vercel AI SDK
ai-sdk.dev

Unified TypeScript SDK for AI apps and agents across model providers.
Agent FrameworksOpen source
Vercel Sandbox
vercel.com

Isolated Linux microVMs on Vercel for untrusted or agent-generated code.
Code Sandboxes
VoidRun
void-run.com

Isolated microVM sandboxes where AI agents and developers run untrusted code, via REST API, TypeScript and Python SDKs, and an MCP server (beta).
Code Sandboxes
Xquik
xquik.com

X (Twitter) data API and MCP server for searching tweets, profiles and followers. Usage-based billing; independent of X Corp.
Search & Web DataOpen source
Zep
getzep.com

Context layer that unifies business data, documents, and chats for agents.
Memory & Context

FAQ

Questions

What is Agent Stack Directory?

A free, checked directory of 84 infrastructure tools for AI agents, grouped into 13 categories like cloud browsers, code sandboxes, memory, and web search.

How are tools chosen?

Each tool is built for or widely used by people building AI agents. Every link is opened and checked before it is listed, and listings are rechecked regularly.

How do I get my tool listed?

Submit it on the submit page. Free listings are reviewed in order. A $79 fast-track gets a decision within 24 hours and is refunded if we decline. Paid placements are always labeled.

Can I pay for a better ranking or a verified badge?

No. Organic order, comparisons and guides are never sold. Paid placements are separate, labeled, and their links are marked rel=sponsored.