Skip to content
HomeFor Agents
MCP Server + REST API

10,600+ tools
for your AI agents

LLMs have stale training data. They hallucinate pricing and miss new tools. Toolradar gives your agents verified, structured, up-to-date intelligence.

10,600+Tools
7,300+Prices verified weekly
5.1MReviews aggregated
400+Categories

Runs in every MCP client

ClaudeChatGPTGrokClaude CodeCursorWindsurfClineVS Code
Claude Desktop
// You ask Claude:
"Find the best free CRM tools with API access"
// Claude calls Toolradar MCP →
search_tools({
query: "CRM API",
pricing: "free",
sort: "score"
})
// → Structured data: names, scores, pricing, categories

Why Toolradar

LLMs forget. Toolradar remembers.

AI models have a knowledge cutoff, typically 1-2 years behind. They recommend defunct tools and quote outdated prices. Toolradar fills that gap with live, structured data.

20%Re-run September 1, 2026: GPT-5.5, Claude Sonnet 5, and Gemini 3.7 Flash named the correct current price for 20% of 70 verified SaaS tools. When they named a number, 42% of those numbers were wrong. Read the study

Pricing verified weekly

Plans change constantly. We verify pricing pages every week so your agent never quotes a price from 2023.

G2 & Capterra ratings

Aggregated review scores from G2, Capterra, Trustpilot, App Store. Real user sentiment, not LLM-generated summaries.

Funding & acquisitions

Latest fundraising rounds, amounts, investors, and acquisitions, tracked via Signalbase. Know which tools are growing and which got acquired.

Structured data

TL;DRs, pros, cons, feature lists, editorial scores, all as JSON. Ready for agent consumption, no parsing needed.

New tools daily

We track Product Hunt, Hacker News, Techpresso, and community submissions. Your agent knows about tools that launched this week.

Real alternatives

AI-identified competitors, not just "same category" matches. Figma's alternatives are Sketch and Adobe XD, not Canva.

Capabilities

What your agents can do

Each tool solves one thing an LLM cannot do well from memory. Here is what each is for.

recommend_tools

Answer “which tool should I use” without guessing

This is the job an LLM cannot fake. Describe the goal and the constraints that actually matter (team size, budget, must-have integrations) and get a ranked shortlist where every pick is justified against the need, with its verified starting price and the trade-offs spelled out. No hallucinated fit, no dead products.

recommend_tools
recommend_tools({
  need: "CRM for a 6-person agency
    that syncs with Gmail",
  max_monthly_budget: 40
})
Returns a fit-scored shortlist, each pick with why it fits, caveats, verified price, and suggested next steps.
get_pricing

Never quote a 2023 price again

Pricing is the fact models get wrong most. We re-check vendor pricing pages every week and hand your agent the current figure with the plan name and the exact date we verified it. In our September 2026 re-run, three current assistants named the correct starting price 20% of the time from memory.

In ChatGPT it renders as an inline pricing card, not a wall of text.

get_pricing
get_pricing({ slug: "figma" })
Returns tiers, the cheapest paid price, a plain-English verdict, and lastVerifiedAt.
compare_tools

Settle a shortlist side by side

Give it 2 to 4 tools and it lays them out on the axes buyers care about (starting price, free tier, free trial) and computes a top pick and a best-value pick. Your agent stops writing vague comparison prose and starts handing over a decision.

Renders as an inline comparison table in ChatGPT.

compare_tools
compare_tools({
  slugs: ["notion", "coda", "slite"]
})
Returns a per-tool table plus a computed Top pick and Best value.
get_alternatives

Find real replacements, not category neighbors

Ask a model for alternatives and it lists whatever shares a tag. We return AI-identified direct competitors: the tools that solve the same job and could actually replace it. Figma's alternatives are Sketch and Adobe XD, not Canva.

get_alternatives
get_alternatives({ slug: "zapier" })
Returns direct competitors ranked by score, each with pricing and a one-line pitch.
search_tools + get_tool

Discover across 10,000+ tools, then drill in

search_tools is Meilisearch over the whole catalog, filterable by category, pricing model, and rating. Found a candidate? get_tool returns the full profile (pros, cons, features, review synthesis, aggregated G2 and Capterra ratings) as clean JSON your agent can reason over.

search_tools + get_tool
search_tools({
  query: "email marketing",
  pricing: "freemium", sort: "rating"
})
Returns ranked matches; then get_tool(slug) for the deep profile.
report_issue

Your agent makes the data better

When an agent notices our data disagrees with the vendor's own page, it files the discrepancy back to our verification queue in one call. Every correction feeds the weekly re-check, so the source gets sharper the more agents lean on it.

report_issue
report_issue({
  slug: "notion", field: "pricing",
  claim: "site shows $10, you show $8"
})
Returns a ticket id and a 48h verification SLA.

Get started

Add it in 30 seconds

1

Paste the URL

Claude, ChatGPT, or Grok → custom connector. No install, no key.

2

Ask one of these

Three prompts that actually call the server. Copy one and send it.

3

Optional: a key

Anonymous is 40 calls/day. A free key is 1,000. Needed only for npx / Cursor.

Remote connector, no install

Claude: Settings → Connectors → Add. ChatGPT: Developer Mode → create a connector. Grok: grok.com/connectors → Custom. Paste:

https://toolradar.com/api/mcp

Then ask one of these

recommend_tools

Recommend a CRM for a 6-person agency under $40/seat that syncs with Gmail.

get_pricing

How much does Figma cost right now, and when was that price last verified?

compare_tools

Compare Notion, Asana, and ClickUp. Who is the top pick and who is the best value?

Or install locally (Cursor, VS Code, Claude Desktop)

claude_desktop_config.json
{
  "mcpServers": {
    "toolradar": {
      "command": "npx",
      "args": ["-y", "toolradar-mcp"],
      "env": {
        "TOOLRADAR_API_KEY": "tr_live_your_key_here"
      }
    }
  }
}

Or via Claude Code: claude mcp add toolradar -- npx -y toolradar-mcp

One-click install

Add to Cursor Add to VS Codeor paste the config into Windsurf, Cline, Zed, or any MCP client.

Local install needs a free key. Get one here. Also on the official MCP registry.

Bonus

/recommend-tool skill

A ready-made Claude Code skill that turns Toolradar into a structured recommendation engine. It searches, compares the top 3, and gives a formatted top pick + runner-up + budget pick.

# Install the skill
claude skill add --url https://raw.githubusercontent.com/Nadeus/toolradar-mcp/main/skill-recommend-tool.md

# Then use it
/recommend-tool "best free CRM for a 5-person startup"

Requires the Toolradar MCP server to be connected. The skill tells Claude how to use the tools for optimal recommendations.

Tools

8 tools, one MCP server

Structured JSON optimized for agent consumption. No HTML, no markdown bloat.

recommend_tools

Justified, sourced shortlist for a described need, ranked by fit with hard budget and pricing constraints

Try: "CRM for a 6-person agency under $40/seat that syncs with Gmail"

search_tools

Search tools by keyword, category, pricing

Try: "free project management tools"

get_tool

Full details on any tool

Try: "Tell me about Linear"

compare_tools

Side-by-side comparison of 2-4 tools

Try: "Notion vs Clickup vs Asana"

get_alternatives

Find alternatives to a tool

Try: "Alternatives to Jira"

get_pricing

Detailed pricing breakdown

Try: "How much does Figma cost?"

list_categories

Browse all categories with tool counts

Try: "What categories exist?"

report_issue

Flag a wrong price, dead tool, or missing feature back to our verification queue

Try: "Notion's site says $10, you show $8"

Pricing

Free to use

No credit card. No commitment. Generate a key and start building.

$0

Free forever

  • 1000 API calls/day
  • All 8 MCP tools
  • Structured JSON responses
  • Updated daily
Get your free key

Are you a tool vendor?

List your tool for free, indexed by Claude, ChatGPT, Perplexity and Google.

Learn more

Beyond the API

Browse the catalogue directly

Frequently asked questions

What is MCP?

MCP (Model Context Protocol) is an open standard that lets AI assistants connect to external data sources and tools. It's supported by Claude Desktop, Claude Code, Cursor, Windsurf, and a growing number of AI clients.

Which AI assistants support this?

Any MCP-compatible client works. This includes Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, Cline, and more. Toolradar is published on the official MCP registry (registry.modelcontextprotocol.io), which feeds the VS Code and GitHub Copilot galleries. You can also use the REST API directly from any HTTP client or agent framework.

How much does an LLM get wrong on its own?

We re-ran the test on September 1, 2026 with GPT-5.5, Claude Sonnet 5, and Gemini 3.7 Flash on the cheapest paid plan of 70 well-known SaaS tools, using prices we had checked the same week. They named the correct price 20% of the time. When they did name a number, 42% of those numbers were wrong. The MCP returns the verified figure with the date we last checked it. See /reports/ai-pricing-knowledge-gap-2026.

How fresh is the data?

Tool data is updated daily. Pricing is verified weekly. New tools are added daily from multiple sources including Product Hunt, Hacker News, and community submissions.

Can I use the REST API directly?

Yes. The MCP server is a thin client over our REST API. You can call the API directly from any language with HTTP support. All endpoints are at toolradar.com/api/v1/.

What are the rate limits?

The free tier includes 1000 API calls per day across all tools. Agents can self-provision a key with no signup by POSTing to /api/v1/keys; anonymous access (no key) is 40/day. Need more? Contact us.

Stop letting agents guess.
Give them Toolradar.

10,600+ tools. Verified pricing. G2 ratings. Updated daily.