agentkey.app

mx

aspmx.l.google.com

alt2.aspmx.l.google.com

alt1.aspmx.l.google.com

alt4.aspmx.l.google.com

alt3.aspmx.l.google.com

ns

charles.ns.cloudflare.com

carol.ns.cloudflare.com

caa

txt

google-site-verification=ctFmFy0FbLL6RSeH2F8ImgMDLrxC0u9PPj1N36ARpbU

v=spf1 include:_spf.google.com ~all

dmarc

User-agent: *
Allow: /
# Install / uninstall scripts are operational assets curl'd from the CLI
# bootstrap flow — they're not pages and shouldn't appear in search
# results (also avoids look-alike domains piggy-backing on the canonical
# install URL via Google searches). Crawlers can still discover them
# from the README / docs that link to them.
#
# The blanket User-agent: * block already covers GPTBot / ClaudeBot /
# PerplexityBot / Google-Extended: every bot we want to allow is
# allowed by default ("Allow: /"), and every bot is denied the four
# script paths below. No need for per-bot sections — those would
# actually defeat the Disallow because each User-agent block is
# independent under the robots.txt spec (the AI-bot blocks here used
# to re-Allow / without re-Disallow'ing the scripts, leaking them).
Disallow: /install.sh
Disallow: /install.ps1
Disallow: /uninstall.sh
Disallow: /uninstall.ps1

Sitemap: https://agentkey.app/sitemap.xml
# Separate sitemap for the /integrations/ landing set, kept out of the main
# sitemap.xml to keep that file hand-maintainable. Generated (and drift-checked
# at build time) by scripts/gen-integrations-sitemap.mjs.
Sitemap: https://agentkey.app/integrations-sitemap.xml
# Ghost serves /blog and generates its own sitemap index (pages / posts /
# authors / tags) — it is not built by build.sh and cannot be merged into
# sitemap.xml. Our sitemap.xml lists only the blog homepage, so without
# this line individual posts are discoverable by crawling alone.
Sitemap: https://agentkey.app/blog/sitemap.xml

# llms.txt is the emerging convention (llmstxt.org) for LLM-driven
# crawlers — distinct from sitemap.xml (pages-for-Google), this is a
# curated content brief for retrieval-augmented agents like ChatGPT
# Search, Perplexity, Claude. robots.txt parsers ignore this comment;
# we document it here so anyone editing the file knows it exists.
# llms-txt: https://agentkey.app/llms.txt

# AgentKey

> AgentKey is a unified MCP (Model Context Protocol) server and API
> gateway that gives AI agents — Claude Code, Cursor, Windsurf, and
> any MCP-compatible client — live access to web search, web extraction,
> social-media data (X, Reddit, YouTube, LinkedIn, TikTok, Instagram),
> and finance / market data. One install, one credit balance, automatic
> provider failover, simple subscription pricing.

AgentKey replaces the "sign up for 8 different paid APIs and wire each
to your agent" workflow with a single key. Built by Chainbase for solo
developers and small AI startups (the "research agent" persona).

## Install

- [Install (one-line)](https://agentkey.app/#setup): `npx @agentkey/cli --auth-login`
- [Console (sign up + get a key)](https://console.chainbase.com/)
- [Install script (POSIX shell)](https://agentkey.app/install.sh)
- [Install script (PowerShell)](https://agentkey.app/install.ps1)

## Docs and code

- [Documentation](https://docs.agentkey.app) — guides, API reference, integration walkthroughs
- [GitHub: AgentKey (server)](https://github.com/chainbase-labs/AgentKey)
- [GitHub: AgentKey Website (this landing)](https://github.com/chainbase-labs/AgentKey-Website)

## What an AI agent gets through AgentKey

- **Web search** — Tavily, Brave, Perplexity, Serper (auto-routed; cheapest first, automatic failover)
- **Web extraction** — Firecrawl, Jina Reader (URL → clean markdown)
- **Social media** — ~46 platforms / ~990 endpoints via TikHub: X (Twitter), Reddit, YouTube, LinkedIn, TikTok, Instagram, Xiaohongshu, Douyin, Bilibili, Weibo, Threads, and more
- **Finance / market data** — Tushare, Yahoo Finance, FRED, Finnhub, Alpha Vantage

## Pricing

Subscription plans, billed monthly or annually (annual saves up to 25%).
One monthly credit balance covers every category — search, extract, social,
and finance. Credits are consumed per call, and different tools cost
different amounts.

- Free — complimentary credits on signup, no card required, never
  auto-converts to a paid plan
- Lite — $12.90/mo (or $9.90/mo billed annually), 500 credits/mo
- Pro — $39.90/mo (or $29.90/mo billed annually), 1,500 credits/mo
- Max — $129.90/mo (or $99.90/mo billed annually), 6,000 credits/mo

Paid subscriptions renew automatically until cancelled; cancel anytime
from the dashboard. Paid subscribers can buy one-time extra credits when
monthly credits run out. Prices are in USD and exclude taxes. See
[Pricing](https://agentkey.app/pricing.html),
[Terms](https://agentkey.app/terms.html) (billing terms in section 4),
and [Privacy](https://agentkey.app/privacy.html).

## Pages

- [Home](https://agentkey.app/) — product overview, features, FAQ
- [Documentation](https://docs.agentkey.app) — guides + API reference
- [Terms of Service](https://agentkey.app/terms.html)
- [Privacy Policy](https://agentkey.app/privacy.html)

## How to use this file

This is an [llms.txt](https://llmstxt.org) — a structured pointer for
LLM-driven crawlers (ChatGPT, Claude, Perplexity, etc.) that prefers
small, well-curated content over raw HTML. The canonical product
description lives on the home page (rendered HTML + JSON-LD); this
file is a fast-path index.