Agent infrastructure
Sandboxes, runtimes, schedulers, MCP hosting and control planes for AI agents. 6 resources; API, key, license and embedding quoted from each site's own pages, checked Oct 24, 2026.
ClawTick
Your agent can schedule its own recurring work with one CLI or API call instead of you keeping a server with crontab, retry code and logging alive.
APIFree tierCommercial OK
InstaVM
Your agent can run untrusted or generated code in its own VM with its own kernel, so a bad script or a prompt injection never touches your machine or your API keys.
APIFree tierCommercial OK
Toolport
You stop copying the same MCP server config and plain-text API keys into every agent's settings file, and a broken server is retried or flagged in one place instead of failing silently in six tools.
Open sourceFree tierCommercial OK
CoreSpeed
Your agent can act in the apps you already use without you registering an OAuth app or handing it API keys for each service, and its memory and connections carry over when you switch agents.
APIFree tierCommercial OK
MCP-Builder.ai
It saves writing, securing and hosting your own MCP server just to let an AI client read or update an internal database, ERP or legacy API.
Free tierCommercial: conditions
Opengeni
Long agent jobs keep running after your laptop sleeps, and API keys sit in a server-side vault that is filled in only when a tool runs, so they never land in .env files or in the model's context.
Open sourceFree tierCommercial OK
Side by side
| Resource | API | Key | Data download | Commercial use | Embed |
|---|---|---|---|---|---|
| ClawTick A hosted cron service that fires scheduled jobs at your OpenClaw agent gateway or any webhook URL, with retries, failure emails, run history, a dashboard, an npm CLI and a REST API. |
Official API · Free tier | Yes | — | Commercial use OK | Embedding unclear |
| InstaVM A cloud service that starts isolated Linux microVMs for AI agents to run code in, with snapshots, clones, persistent volumes, egress rules, secret injection, public port shares and a browser desktop, driven from a CLI, SSH, Python or TypeScript. |
Official API · Free tier | Yes | — | Commercial use OK | Embedding unclear |
| Toolport A free MIT-licensed desktop app for Windows, macOS and Linux that runs one local MCP gateway: you add each MCP server once, it replaces the per-agent config entries in 39 detected agents (Claude Code, Codex, Cursor and others) with a single entry, and moves API keys into the OS keychain. |
No official API | — | — | Commercial use OK | Embedding unclear |
| CoreSpeed One remote MCP endpoint that gives any agent (Claude Code, Codex, Cursor and others) OAuth-connected tools for apps like Gmail, Slack, GitHub, Notion and HubSpot, shared memory across agents, web search and scraping, media generation, plus spend caps, approval rules and an activity log. |
Official API · Free tier | Yes | — | Commercial use OK | Embedding unclear |
| MCP-Builder.ai A service that generates an MCP server from a plain-language description, an OpenAPI/WSDL spec, an XML sample or a SQL schema, then hosts it with API-key or OAuth auth and audit logs so Claude, ChatGPT, Copilot or Cursor can call your APIs and databases through one URL. |
API unclear | — | — | Commercial use with conditions | Embedding unclear |
| Opengeni An open-source platform that gives each AI agent its own cloud machine (git, shell, Python, browser) and lets you start long tasks from a web app, Slack, a schedule or an HTTP API, with a TypeScript SDK and React components to embed the agent in your own product. |
API unclear | — | — | Commercial use OK | Can be embedded |