Forency
# Forency MCP Server
[](https://smithery.ai/servers/nathan-cheever12/forency-mcp)
Let your AI assistant investigate any website's technology stack — CMS, ecommerce platform, JavaScript frameworks, analytics, CRM, marketing automation, payments, chat widgets, CDN, and hosting.
Ask Claude (or Cursor, or any MCP client):
> "What tech stack does competitor.com run?"
> "Which of these 15 prospects use HubSpot vs Salesforce?"
> "Compare the analytics and payment stacks of these three stores."
## Two ways to connect
**Hosted — no install.** If your client supports remote MCP servers, point it at
`https://forency.io/api/mcp` with your API key as a Bearer token. Nothing to
download, and you always get the current tools. Also listed on
[Smithery](https://smithery.ai/servers/nathan-cheever12/forency-mcp).
**Local — this package.** For clients that run MCP servers on your machine:
## Install
Add to your MCP config — `claude_desktop_config.json`, `.mcp.json`, or your client's equivalent:
```json
{
"mcpServers": {
"forency": {
"command": "npx",
"args": ["-y", "forency-mcp"],
"env": { "FORENCY_API_KEY": "fcy_your_key" }
}
}
}
```
Or with Claude Code:
```bash
claude mcp add forency -e FORENCY_API_KEY=fcy_your_key -- npx -y forency-mcp
```
Get a free API key at [forency.io](https://forency.io) — 100 scans/month, no credit card.
## Tools
| Tool | What it does |
|------|--------------|
| `lookup_tech_stack(url)` | Full technology breakdown for one website |
| `batch_lookup(urls)` | Up to 25 websites in one call, full detail for each |
| `compare_stacks(urls, categories?)` | Side-by-side table across sites, optionally filtered to categories like `["CRM", "Analytics"]` |
Each URL scanned counts as one scan against your plan.
## What this package does, and doesn't
This repository is **the whole client**. It is a thin stdio MCP server that forwards your request to the Forency REST API and formats the reply. Worth knowing before you hand any tool an API key:
- it talks to `https://forency.io` and nothing else
- it sends your API key and the URLs you asked about, nothing more
- no telemetry, no analytics, no third-party calls
- detection happens server-side, so it is not in this repo
You can read the entire thing in [`src/index.ts`](src/index.ts) in about five minutes. Point it somewhere else with `FORENCY_API_URL` if you want to watch what it sends.
## Environment
| Variable | Required | Description |
|----------|----------|-------------|
| `FORENCY_API_KEY` | Yes | Your key from [forency.io](https://forency.io) |
| `FORENCY_API_URL` | No | API base URL (defaults to `https://forency.io`) |
## A note on detection
Forency reads the HTML and headers a site's server returns. It does not run a
browser, so technologies injected purely by client-side JavaScript may not show
up — a sparse result means little was *detectable*, not that a site runs nothing.
## Pricing
| Tier | Price | Scans/mo |
|------|-------|----------|
| Free | $0 | 100 |
| Starter | $29/mo | 1,000 |
| Pro | $49/mo | 10,000 |
| Enterprise | Custom | Unlimited |
BuiltWith's API starts at $295/mo and Wappalyzer's at $250/mo. Neither ships an MCP server.
## Acknowledgements
Forency's detection is built on the technology fingerprints maintained by the
[**webappanalyzer**](https://github.com/enthec/webappanalyzer) project, the
community-maintained continuation of Wappalyzer's open-source fingerprint
database, which is licensed **GPL-3.0**. That work is what makes accurate
detection possible at this price, and it deserves the credit.
This MCP client contains no fingerprint data or detection code of its own and is
released under the MIT license (see [LICENSE](LICENSE)).
## License
MIT
TDQS
Scored across 3 tools
Each tool serves a distinct workflow: single-site lookup, bulk detail lookup, and side-by-side comparison. The descriptions explicitly cross-reference each other, making it clear when to choose one over the other.
Two names follow a verb_noun pattern (lookup_tech_stack, compare_stacks), while batch_lookup is a modifier_noun form. The inconsistency is minor and all names are still clear and predictable.
Three tools is well-scoped for a tech-stack detection server. Each tool covers a distinct usage size: single, batch, and comparative, with no redundant or unnecessary additions.
The tool surface covers the full intended workflow: looking up one site, looking up many sites, and comparing stacks across sites. There are no obvious dead ends or missing operations within the stated domain.