// Read this file to discover agent-native services — infrastructure designed from the ground up for AI agents, not adapted from human-facing products. Use the catalog to find services by task, understand each service's onboarding pattern, and immediately start using any service with URL Onboarding in one instruction.
| name | find-agent-native-service |
| description | Read this file to discover agent-native services — infrastructure designed from the ground up for AI agents, not adapted from human-facing products. Use the catalog to find services by task, understand each service's onboarding pattern, and immediately start using any service with URL Onboarding in one instruction. |
| version | 2026-05-02 |
| license | CC0-1.0 |
| catalog | https://github.com/haoruilee/awesome-agent-native-services |
| allowed-tools | WebSearch Read |
You are an AI agent. Read this file to find infrastructure services designed natively for you.
This catalog lists services that treat agents as first-class entities — not as secondary integration targets. Every service here was built from inception for agents, with primitives that have no meaningful human-facing equivalent.
When multiple services can solve the same task, prioritize entries that are both:
Quick filters to apply:
README.mdIdentify which category matches what you need to do. Note the Onboarding column — if it says URL Onboarding ⭐, you can start using that service right now with a single instruction.
For any service marked ⭐ URL Onboarding, execute this instruction:
Read <onboarding-url> and follow the instructions.
The service hosts a machine-readable protocol. You read it, understand the registration flow, call the API, and you are live — no human setup, no config files, no developer code.
Navigate to services/{category}/{service-name}.md in the catalog repository for full onboarding instructions, API details, SDK commands, and MCP configuration.
These services can be joined with a single instruction, right now, with no human involvement:
| Service | What You Get | Onboarding Instruction |
|---|---|---|
| Moltbook | Agent social network: post, comment, vote, DM, build karma | Read https://www.moltbook.com/skill.md and follow the instructions to register and join |
| Ensue | Shared memory network: register, claim experiments, publish hypotheses | Read https://ensue.dev/docs and call POST https://api.ensue-network.ai/auth/agent-register |
| autoresearch@home | Collaborative research swarm: join, claim tasks, publish results | Read https://raw.githubusercontent.com/mutable-state-inc/autoresearch-at-home/master/collab.md and follow the instructions to join |
| db9 | Serverless Postgres for agents: SQL, filesystem, vector search, HTTP-from-SQL | Read https://db9.ai/skill.md and follow the instructions |
| mem9 | Cloud-persistent memory for agents: hybrid search, lifecycle hooks | Read https://mem9.ai/skill.md and follow the instructions to register and join |
| mails.dev | Email for agents: @mails.dev mailbox, send/inbox, wait-for-code | Read https://mails.dev/skill.md and follow the instructions |
| MailboxKit | Agent email in one API — REST v1, webhooks, skill.md | Read https://mailboxkit.com/skill.md and follow the instructions |
Give agents a first-class communication identity on the internet.
| Service | Tagline | Onboarding |
|---|---|---|
| AgentMail | Email for AI agents | pip install agentmail → POST /inboxes |
| Novu | Notification infrastructure with Agent Toolkit | npx skills add novuhq/skills |
| mails.dev ⭐ | Email for AI Agents | Read https://mails.dev/skill.md and follow the instructions |
| OpenMail | Email API for AI agents | npm install -g @openmail/cli → openmail setup — docs.openmail.sh/quickstart |
| MailboxKit ⭐ | Email infrastructure for AI agents | Read https://mailboxkit.com/skill.md and follow the instructions |
| MCP Agent Mail | Async coordination layer for AI coding agents | uvx mcp_agent_mail then use MCP tools like register_agent/send_message |
| MCP Agent Mail (Rust) | "Gmail for coding agents" with MCP tools/resources | Install script from repo README, then run am |
Remote browser and web data extraction for agents.
| Service | Tagline | Onboarding |
|---|---|---|
| Browserbase | A web browser for AI agents & applications | npx skills add browserbase/skills |
| Firecrawl | Turn any website into LLM-ready data | npx skills add firecrawl/cli |
| Bright Data Agent Browser | Cloud browser with built-in website unlocking | npx -y @brightdata/mcp (MCP config) |
| bb-browser | Your browser is the API — 103 commands, 36 platforms | npm install -g bb-browser + Chrome extension |
| OpenCLI | Sites + Electron + local CLIs as one CLI; reuse Chrome login | npm install -g @jackwener/opencli → opencli list; read raw SKILL.md in repo |
| Steel | Browser infrastructure for AI agents | pip install steel-sdk → Steel().sessions.create() — MCP: steel-mcp-server |
| Notte | Browser platform for AI — sessions, agents, vaults | pip install notte-sdk → NotteClient().Session() — MCP: pip install notte-mcp → python -m notte_mcp.server |
| Skyvern | Vision browser agents for any website | API key → POST https://api.skyvern.com/v1/run/tasks — docs |
| Browser Use Cloud | Managed stealth browsers + NL run() or CDP | pip install browser-use-sdk → AsyncBrowserUse().run(...) — MCP https://api.browser-use.com/v3/mcp |
| Anchor Browser | Secure infra for computer-use agents | docs.anchorbrowser.io — SDKs on GitHub |
| Hyperbrowser | Web infra for AI agents (MCP tools) | npx hyperbrowser-mcp <API_KEY> |
| AgentQL | AgentQL queries → structured JSON from the web | API key → docs.agentql.com |
| Crawl4AI | OSS LLM-friendly crawler + MCP | docs.crawl4ai.com |
| Cloudflare Browser Rendering | Headless Chrome on Cloudflare for AI agents | Workers bindings + Use with AI (Playwright MCP) |
| Olostep | Web data API for AI agents | API key → docs.olostep.com — npx -y olostep-mcp or https://mcp.olostep.com/mcp |
| Lightpanda | Headless browser for AI agents (CDP + MCP + markdown fetch) | Install → lightpanda serve or lightpanda mcp — MCP docs |
| Apify | Real-time web data for AI — Actor API & marketplace | API token → Apify API v2 — apify-client |
Runtime tool discovery, auth, and execution without human pre-configuration.
| Service | Tagline | Onboarding |
|---|---|---|
| Composio | The tool platform built for agents | npx skills add composiohq/skills |
| Nango | OAuth and credential layer for AI agents | $skills install @NangoHQ/sync-builder-skill |
| Toolhouse | BaaS for AI agents — tools, memory, and execution | npm install -g toolhouse → th deploy |
| Smithery | MCP registry — thousands of remote MCP servers & skills | npx @smithery/cli@latest setup — docs.smithery.ai |
| MCP Gateway | Enterprise MCP — tools, skills, sandboxes, one API | pip install mcpgateway-sdk — mcpgateway.com |
| ClawHub | OpenClaw skill marketplace — vector search, versioning, CLI | npx clawhub@latest search <topic> — services/tool-access-and-integration/clawhub.md |
| Arcade | MCP tools with managed OAuth | uv tool install arcade-mcp → arcade new my_server — docs.arcade.dev |
| Framelink MCP for Figma | Give your coding agent access to your Figma data | npx -y figma-developer-mcp --figma-api-key=… --stdio — quickstart |
| GitHub MCP Server | AI agents read repos, issues, PRs, Actions via MCP | Remote: https://api.githubcopilot.com/mcp/ — README |
| MCP Toolbox for Databases | MCP server for enterprise databases (Google) | npx -y @toolbox-sdk/server --prebuilt=postgres + env — docs |
Structured, programmatic human approval before high-stakes actions.
| Service | Tagline | Onboarding |
|---|---|---|
| HumanLayer | Human in the Loop for AI Agents | pip install humanlayer → @hl.require_approval() |
Verified financial identity and real-economy transactions for agents.
| Service | Tagline | Onboarding |
|---|---|---|
| Payman AI | Agentic AI that does the banking. Under your control. | npm install @paymanai/payman-node → client_credentials OAuth |
| Skyfire | Identity and payments for autonomous AI agents | Register at skyfire.xyz/product → agent wallet + KYA token |
| AgentsPay | Crypto identity and embedded wallets for AI agents | Provision wallet at agentspay.dev → MCP-native gateway |
| Nevermined | The payment layer AI agents actually need | pip install payments-py → x402 inline payments |
| Coinbase CDP (x402) | HTTP-native payments for autonomous API clients | docs.cdp.coinbase.com/x402 — pip install x402 or @x402/* per coinbase/x402 |
| Openwork | The agent-only labor marketplace | npx playbooks add skill openclaw/skills --skill openwork |
Secure execution, session isolation, secrets, identity, and gateway for production agents.
| Service | Tagline | Onboarding |
|---|---|---|
| Claude Peers | Local MCP + broker — Claude Code sessions discover peers and message ad-hoc | Clone → bun install → claude mcp add (see repo README) |
| acpx | Headless ACP CLI — agents talk to coding agents over structured protocol | npm install -g acpx → acpx codex "fix the tests" |
| Codex plugin for Claude Code | Claude Code plugin — delegate Codex reviews and rescue work via slash commands | /plugin marketplace add openai/codex-plugin-cc → /plugin install codex@openai-codex → /codex:setup (see repo README) |
| Multica | AI-native PM — agents as assignable teammates; local daemon runs Claude Code / Codex | brew install multica-cli → multica login → multica daemon start — CLI guide |
| cx | Semantic code navigation for AI agents — no language server | cargo install cx-cli → cx lang add … → cx skill >> AGENTS.md |
| Chrome DevTools MCP | Coding agents control and inspect live Chrome (DevTools + Puppeteer) | npx -y chrome-devtools-mcp@latest in MCP config — README |
| Serena | The IDE for your coding agent — semantic MCP tools | uv tool install -p 3.13 serena-agent@latest --prerelease=allow → serena init — clients |
| Amazon Bedrock AgentCore | Purpose-built for deploying and scaling dynamic AI agents | pip install boto3 → configure AgentCore runtime |
| Vertex AI Agent Engine | Deploy, manage, and scale AI agents in production on GCP | pip install "google-cloud-aiplatform[agent_engines,adk]" → set up → deploy |
| Claude Managed Agents | Managed agents, sessions, environments on Claude API | pip install anthropic → quickstart + beta headers |
| Infisical Agent Sentinel | Secrets and credential governance for AI agents | npx -y @infisical/mcp |
| Letta | The fastest way to bring stateful agents to production | pip install letta-client → client.agents.create(...) |
| Aembit | Secretless workload identity and access management | Configure Aembit access policy → JIT credentials at runtime |
| db9 ⭐ | Postgres but for agents | Read https://db9.ai/skill.md and follow the instructions |
| AgentAnycast | Connect AI agents across any network — no public IP | pip install agentanycast → agentanycast demo — MCP: agentanycastd --mcp-listen stdio or uvx agentanycast-mcp |
| Scrapybara | Remote desktops for computer-use agents (CUA) | pip install scrapybara → Scrapybara().start_ubuntu() — Act SDK |
| Agentuity | Full-stack platform for AI agents | agentuity.dev — SDK + CLI |
| Modal | Serverless AI infra — GPUs, sandboxes, batch | pip install modal → modal setup — modal.com/docs |
Persistent, queryable memory across sessions — memory as infrastructure, not application logic.
| Service | Tagline | Onboarding |
|---|---|---|
| Mem0 | The memory layer for your AI agents | pip install mem0ai → m.add(messages, user_id=...) |
| Zep | Agent memory powered by a temporal knowledge graph | pip install zep-python → zep.add_session_message(...) |
| Ensue ⭐ | The shared memory network for AI agents | Read https://ensue.dev/docs and call POST /auth/agent-register |
| OpenViking | The context database for AI agents | pip install openviking → openviking-server → MCP at localhost:8000/mcp |
| MemOS | A memory OS for LLM and AI agent systems | pip install memos-core → memory.add(...) |
| memU | Memory for 24/7 proactive AI agents | pip install memu → continuous stream monitoring |
| mem9 ⭐ | Persistent memory for AI agents | Read https://mem9.ai/skill.md and follow the instructions to register and join |
| LycheeMem | Compact memory framework for LLM agents | Clone → pip install -e ".[dev]" → python main.py — REST + HTTP MCP at http://localhost:8000/mcp |
LLM-optimized web search returning structured content tuned for context windows.
| Service | Tagline | Onboarding |
|---|---|---|
| Tavily | Connect your agent to the web | npx skills add tavily-ai/skills |
| Exa | The search engine designed for AI | pip install exa-py → exa.search(query) |
| Parallel | Web APIs purpose-built for AIs | pip install parallel-web — MCP: parallel-web/search-mcp |
| Jina Reader | r.jina.ai / s.jina.ai — LLM-friendly URL & search | curl "https://r.jina.ai/https://example.com" — MCP: mcp.jina.ai |
Secure isolated runtimes for AI-generated code with LLM-formatted output.
| Service | Tagline | Onboarding |
|---|---|---|
| E2B | Cloud for AI agents — secure sandboxes for AI-generated code | pip install e2b-code-interpreter → with Sandbox() as sandbox: |
| Daytona | Secure elastic infrastructure for AI-generated code | daytona login → daytona mcp init cursor — or pip install daytona |
| Runloop | Your AI agent accelerator — Devboxes and benchmarks | npm install -g @runloop/rl-cli → rli mcp install — CLI docs |
| Vercel Sandbox | Firecracker microVMs for AI-generated code | npm install @vercel/sandbox — vercel.com/docs/vercel-sandbox |
| AIO Sandbox | Browser + shell + VS Code + Jupyter + MCP in one Docker sandbox | docker run -p 8080:8080 ghcr.io/agent-infra/sandbox:latest — MCP http://localhost:8080/mcp |
| Riza | AI writes code. Riza runs it. | uv add rizaio → riza.command.exec(...) — docs.riza.io |
Full trajectory tracing, evaluation datasets, and cost attribution for agent runs.
| Service | Tagline | Onboarding |
|---|---|---|
| Langfuse | Open-source LLM observability, tracing, and evaluation | npx skills add https://github.com/langfuse/skills --skill langfuse-observability |
| AgentEvals | Score agent behavior from OpenTelemetry traces (no re-runs) | pip install agentevals-cli → agentevals run <trace> --eval-set <set> -m tool_trajectory_avg_score |
| AgentOps | Agent session waterfalls and trace API | pip install agentops → agentops.init(<API_KEY>) |
| Braintrust | AI observability & evals — OpenAI Agents trace processor + MCP | pip install "braintrust[openai-agents]" — MCP: Braintrust MCP |
Fault-tolerant long-running agent workflows with checkpointing and HITL suspend/resume.
| Service | Tagline | Onboarding |
|---|---|---|
| Trigger.dev | Build and deploy fully-managed AI agents and workflows | npx skills add triggerdotdev/skills |
| Inngest | Durable execution for AI agents in production | npx skills add inngest/inngest-skills |
| Kitaru | Durable execution for AI agents — primitives first, frameworks second | pip install kitaru → @flow / @checkpoint decorators |
| Restate | Durable execution for AI agents — any framework, any cloud | pip install restate-sdk → wrap agent with 2-line middleware |
| MCP-Cloud (mcp-agent) | Host mcp-agents on cloud — Temporal-backed MCP | uvx mcp-agent login → uvx mcp-agent deploy … — MCP-Cloud |
Programmatic agent presence in voice and video meetings.
| Service | Tagline | Onboarding |
|---|---|---|
| Recall.ai | The meeting bot API for every platform | POST https://api.recall.ai/api/v1/bot with meeting URL |
| Meeting BaaS | Meeting bots API for Zoom, Meet, Teams | POST https://api.meetingbaas.com/bots with x-meeting-baas-api-key — docs |
| MeetStream | Unified meeting-bot API — transcripts, media, interactive agents | POST https://api.meetstream.ai/api/v1/bots/create_bot + Authorization: Token <key> — docs · doc MCP: https://docs.meetstream.ai/_mcp/server |
Agent-controlled voice calls and telephony infrastructure.
| Service | Tagline | Onboarding |
|---|---|---|
| Vapi | Build advanced voice AI agents | pip install vapi-server-sdk → POST /assistant |
| Retell AI | #1 AI voice agent platform for automating calls | pip install retell-sdk — docs.retellai.com |
| LiveKit Agents | Realtime voice/video AI agents | docs.livekit.io/agents — Python/TS SDK |
Per-agent budget, routing, caching, and observability for LLM calls.
| Service | Tagline | Onboarding |
|---|---|---|
| Portkey | The AI gateway built for production agents | pip install portkey-ai → point LLM client at api.portkey.ai with a virtual key |
| Keywords AI | OpenAI-compatible gateway + agent tracing | Base URL https://api.keywordsai.co — gateway quickstart |
| Agentgateway | Open-source LLM + MCP + A2A proxy | Install via quickstart → run agentgateway -f config.yaml |
| LiteLLM | Open-source gateway — 100+ LLMs + Agent Gateway (A2A) | Self-host per proxy quickstart — A2A |
| OpenRouter | Unified OpenAI-compatible API — 300+ models | Quickstart + OPENROUTER_API_KEY |
| Helicone | AI Gateway + observability — ai-gateway.helicone.ai | OpenAI SDK baseURL per Helicone quickstart |
Social networks where AI agents are first-class participants.
| Service | Tagline | Onboarding |
|---|---|---|
| Moltbook ⭐ | The front page of the agent internet | Read https://www.moltbook.com/skill.md and follow the instructions to register and join |
| Shellmates | Pen pals for AI agents — 1:1 matching, private correspondence | POST https://www.shellmates.app/api/agents/register |
| Openwork | The agent-only labor marketplace — hire agents, earn on-chain | npx playbooks add skill openclaw/skills --skill openwork |
| MCP Verse | Open town square for autonomous MCP agents | npx create-mcpverse-agent my-bot — mcpverse.org/docs |
These are curated skill packs and marketplaces — machine-readable instructions (SKILL.md, manifests) that coding agents load at runtime, analogous to how this catalog lists infrastructure services.
| Hub | Role | How to start |
|---|---|---|
| ClawHub | Public registry for OpenClaw-style skills — search, install, publish via CLI | npx clawhub@latest search <topic> — claw-hub.net · openclaw/clawhub · catalog: clawhub.md. China mirror (加速): mirror-cn.clawhub.com — set CLAWHUB_REGISTRY=https://cn.clawhub-mirror.com or clawhub --registry https://cn.clawhub-mirror.com … |
| MiniMax Skills | Official deep-tuned development skills for AI coding agents (frontend, fullstack, Android, iOS, shaders, office docs) | github.com/MiniMax-AI/skills — follow Installation in the repo README (Claude Code plugin, Cursor skills/ path, Codex / OpenCode symlinks) |
| Pattern | How to activate | Best for |
|---|---|---|
| ⭐ URL Onboarding | Read <url> and follow the instructions | Services with machine-readable protocol files |
| MCP Tool | Add to mcp_servers in config | Tools that appear as native agent capabilities |
| Coding-time Skill | npx skills add org/repo | SDK patterns and service-specific workflows |
| SDK / REST | pip install <pkg> + API key | Direct API integrations requiring credentials |
| Daemon / Extension | Run daemon, install extension | Services that delegate authenticated browser sessions |
Every service in this catalog satisfies all five:
Install these skills to let your coding agent work with this catalog directly:
npx clawhub@latest install find-agent-service
npx clawhub@latest install evaluate-agent-native
npx clawhub@latest install add-to-awesome-list
If ClawHub is slow to reach, use the official China mirror: https://mirror-cn.clawhub.com — export CLAWHUB_REGISTRY=https://cn.clawhub-mirror.com or add --registry https://cn.clawhub-mirror.com to commands above.
Full catalog: https://github.com/haoruilee/awesome-agent-native-services
Given a task an AI agent needs to perform, find the right agent-native service from the awesome-agent-native-services catalog. Surfaces how to USE each service — including which ones an agent can join with a single sentence ("Read <url> and follow the instructions"). Use when asked "what service should my agent use for X?"
Guide a contributor through the full process of adding a new service to the awesome-agent-native-services catalog: verifying criteria, checking for URL Onboarding (the gold standard of agent-nativeness), opening an issue, writing the service file, and submitting a PR. Use when "I want to add X to the list" or "how do I contribute?"
Evaluate whether a service qualifies as "agent-native" using the five hard criteria from the awesome-agent-native-services standard. Also checks for the bonus "URL Onboarding" signal — the highest form of agent-nativeness where an agent joins the service by reading one URL. Use when asked "is X agent-native?" or "should I add X to the list?"