@genready/mcp-server
# @genready/mcp-server
Connect [GenReady](https://genready.ai) to your AI assistant or coding agent. Scan any page for readiness to be cited by AI search engines and used by AI agents, then hand the fix checklist straight to your agent.
## 1. Get an API key
Create one at [genready.ai/api-keys](https://genready.ai/api-keys). To read site-wide fix plans, enable **Read site data** (`domain:read`) when you create it.
## 2. Connect
### Remote (recommended, nothing to install)
Any client that supports remote MCP servers over Streamable HTTP can connect directly:
- URL: `https://genready.ai/mcp`
- Header: `Authorization: Bearer gr_live_your_key_here`
Claude Code:
```bash
claude mcp add --transport http genready https://genready.ai/mcp \
--header "Authorization: Bearer gr_live_your_key_here"
```
Cursor (`.cursor/mcp.json`):
```json
{
"mcpServers": {
"genready": {
"url": "https://genready.ai/mcp",
"headers": { "Authorization": "Bearer gr_live_your_key_here" }
}
}
}
```
### Local command (stdio)
For clients that only launch local commands, such as Claude Desktop:
```json
{
"mcpServers": {
"genready": {
"command": "npx",
"args": ["-y", "@genready/mcp-server"],
"env": { "GENREADY_API_KEY": "gr_live_your_key_here" }
}
}
}
```
This package forwards to the hosted endpoint, so both options expose the same tools.
## 3. Ask
> Scan https://example.com for AI readiness and fix what you can in this repo.
## Tools
| Tool | What it does | Credits |
| --------------------- | --------------------------------------------------------------------------------------------- | ---------- |
| `analyze_website` | Scan a URL. Returns scores and the fix checklist. `scope`: `full`, `content`, `crawlability`. | 1 per scan |
| `get_report` | Read a stored report by ID, including its fix checklist. | Free |
| `check_status` | Progress of a scan that did not finish, or was started with `wait: false`. | Free |
| `list_reports` | Your past reports, with filters for URL and status. | Free |
| `get_findings` | The fix list for a page or a site, with stable ids. | Free |
| `verify_findings` | Re-check findings against the live site after you deploy a fix (Pro and Agency). | Free |
| `update_finding` | Record that a finding is fixed, pending, or ignored. | Free |
| `get_artifact` | Generate a `robots.txt`, `llms.txt` or JSON-LD file that fixes a finding. | Free |
| `start_site_analysis` | Scan a whole site (needs `site:analyze`). Capped at 25 pages over OAuth. | 1 per page |
| `get_usage` | Plan, credits used and remaining. | Free |
| `list_domains` | Your monitored domains (needs `domain:read`). | Free |
| `get_fix_plan` | Ranked site-wide fixes from a domain's latest site analysis (needs `domain:read`). | Free |
The fix checklist is included on Pro and Agency plans. On other plans, reports say it is locked.
## Environment variables (stdio)
| Variable | Required | Description |
| ------------------ | -------- | ------------------------------------------------------- |
| `GENREADY_API_KEY` | Yes | Your API key (starts with `gr_live_`) |
| `GENREADY_MCP_URL` | No | Hosted MCP endpoint (default `https://genready.ai/mcp`) |
## Links
- [MCP docs](https://genready.ai/docs/mcp)
- [REST API docs](https://genready.ai/docs/api)
## Support and contributing
- Found a bug or a wrong tool description? [Open an issue](https://github.com/odb366/genready-mcp-server/issues/new). Include the tool name, what you asked, and the error text.
- Questions about your account or plan: [genready.ai/support](https://genready.ai/support).
- This repository is a published copy of the MCP server that lives in the GenReady monorepo. See [CONTRIBUTING.md](CONTRIBUTING.md) before you open a pull request.
TDQS
Scored across 12 tools
Most tools have clearly distinct roles: scanning (analyze_website, start_site_analysis), reading (get_report, get_findings, get_fix_plan), and lifecycle (update_finding, verify_findings, get_artifact). The get_report/get_findings/get_fix_plan trio overlaps somewhat, but descriptions explicitly steer ('use get_report instead if you only need scores'), and the page-vs-site distinction is stated.
All twelve tools use consistent snake_case verb_noun form (get_usage, list_reports, check_status, start_site_analysis, verify_findings). No mixing of camelCase or vague single-verb names.
Twelve tools falls squarely in the well-scoped 3-15 range, and each maps to a distinct step in the scan-fix-verify workflow. No redundant or filler tools.
The surface covers the full workflow: credit check, single/site scanning, status polling, report and findings reads, artifact generation, finding state updates, and post-deploy verification. Minor gaps exist around domain management (only list_domains, no add/remove) and report cleanup, but core CRUD/lifecycle is covered.