wcagc-mcp
# wcagc-mcp
[](https://glama.ai/mcp/servers/WCAG-Compliance/wcagc-mcp)
[](https://smithery.ai/servers/wcag-compliance/wcagc-mcp)
An [MCP](https://modelcontextprotocol.io) server that lets an AI assistant (Claude, ChatGPT, or
any MCP-compatible client) run real, deterministic accessibility scans through
[wcagc](https://wcagc.com) — axe-core under the hood, not an LLM guess. Every scan result carries
an explicit coverage disclaimer and never claims "compliant": automated testing finds only a
portion of accessibility barriers, and this tool says so in every response.
This package is a **thin, stateless adapter**. It holds no database, no scan logic, and no
secrets beyond the wcagc API base URL — it translates MCP tool calls into HTTP calls against the
wcagc API and forwards the caller's own bearer. All authentication, entitlements, quotas, and
scan orchestration live in the API; this code is safe to read end to end.
## Two ways to run it
**Local (stdio)** — for Claude Desktop, Cursor, or any MCP client that spawns a local process:
```bash
npx @wcagc/mcp
```
Configure your MCP client with:
```json
{
"mcpServers": {
"wcagc": {
"command": "npx",
"args": ["-y", "@wcagc/mcp"],
"env": {
"WCAGC_MCP_KEY": "<your mcp:scan API key>"
}
}
}
}
```
Mint an `mcp:scan` key from your wcagc account under Settings → API keys — available on every
plan, with a daily quota on Free/Starter and unlimited on Pro/Agency.
**Hosted (Streamable HTTP + managed OAuth)** — what Claude web/desktop/mobile connectors and
ChatGPT use, since neither runs a local process for you. Add this remote MCP connector:
```
https://mcp.wcagc.com/mcp
```
The client discovers `/.well-known/oauth-protected-resource/mcp`, opens the wcagc login/consent
flow, and binds the connection to one Organization. No key copy/paste is required. API-key bearer
authentication remains supported for local stdio and CI.
ChatGPT availability depends on the ChatGPT plan and on whether the client permits action tools;
`scan_url` creates a scan and is not a read-only operation. See
[wcagc.com/integrations/mcp](https://wcagc.com/integrations/mcp).
## Tools
| Tool | Plan | What it does |
|---|---|---|
| `scan_url` | all | Scan any public URL, or a registered site for full tracking (Pro+). |
| `check_pdf` | all | Run a PDF/UA-1 structure check on a public PDF. |
| `get_scan` · `get_findings` | all | Read a `scan_url` scan's status, severity counts, and findings — either kind, recorded or one-off. |
| `list_sites` | Pro+ | List the account's registered sites. |
| `scan_site` | Pro+ | Crawl and scan every reachable page of a registered site. |
| `get_run` · `get_run_findings` | Pro+ | Read a full-site run from `scan_site`. |
| `get_root_causes` | Pro+ | Group a run's repeated DOM patterns and return factual element/page blast radius. |
| `run_journey` | Pro+ | Replay a saved multi-step journey and check each step. |
| `get_trends` | Pro+ | Read a site's violation-count history over time. |
One id, one poll tool: whatever `scan_url` did with a URL, `get_scan` and `get_findings` read it
back. `get_run` and `get_run_findings` are only for full-site runs from `scan_site`.
Every scan-producing tool returns the coverage disclaimer in both the text content and the
structured content. There is no score, grade, or conformance verdict — automated testing finds
roughly 30–57% of accessibility issues, and the remainder needs manual review.
## Configuration
| Env var | Used by | Meaning |
|---|---|---|
| `WCAGC_API_BASE_URL` | both | The wcagc API to call. Defaults to `https://api.wcagc.com`; set it only when self-hosting. |
| `WCAGC_MCP_KEY` | stdio | Your `mcp:scan` API key. |
| `PORT` | hosted | Port to listen on (default `8080`). |
| `WCAGC_MCP_ALLOWED_HOSTS` | hosted | Comma-separated Host-header allowlist (DNS-rebinding protection when bound to `0.0.0.0`). |
| `WCAGC_MCP_INTROSPECT_TTL_SECONDS` | hosted | How long a verified bearer is cached before re-checking with the API (default `60`). |
| `WCAGC_MCP_OAUTH_ISSUER` | hosted | Expected OAuth issuer (defaults to `WCAGC_API_BASE_URL`). |
| `WCAGC_MCP_OAUTH_JWKS_URL` | hosted | Authorization Server public JWKS URL. |
| `WCAGC_MCP_URL` | hosted | Canonical RFC 9728 protected-resource URL (defaults to `https://mcp.wcagc.com/mcp`). |
| `WCAGC_MCP_JWKS_TTL_SECONDS` | hosted | JWKS cache TTL; an unknown `kid` triggers an immediate refetch. |
| `WCAGC_OPENAI_APPS_CHALLENGE` | hosted | OpenAI Plugins Directory domain-verification token; keep it in the deployment secret store, never in source. |
## Development
```bash
npm install
npm run dev # hosted, watch mode
npm run start:stdio # stdio mode
npm run typecheck
npm run verify # node:test against a local fixture API
```
## License
MIT — see [LICENSE](./LICENSE).
TDQS
Scored across 16 tools
Most tools are clearly scoped to distinct resources and actions: scan_url/scan_site, check_pdf, run_journey, and verify_fix each have their own dedicated polling tool. The many get_* tools could be confused at a glance, but descriptions explicitly cross-reference which poller to use for each job type.
All tools follow a consistent lowercase snake_case verb_noun pattern (scan_url, list_sites, get_run_findings, verify_fix). Paired queue/poll tools are predictably named (check_pdf -> get_pdf_check, run_journey -> get_journey_run), making the set easy to navigate.
16 tools is slightly above the ideal 3-15 range but each tool maps to a distinct stage in the scanning pipeline: initiating a job, polling it, retrieving findings, and listing supporting resources. The count feels justified by the breadth of scan types rather than redundant.
The surface covers the core lifecycle for URL scans, full-site runs, PDF checks, journeys, trend history, and fix verification. Minor gaps exist — no site registration/update/delete and no cancellation or re-run operations — but these are likely managed outside the MCP and don't block primary workflows.