Skip to main content
Glama

Webtrax runner

The open source engine behind Webtrax, the agentic platform for website quality. It crawls a website once and checks it for broken links, redirect problems, missing security headers, SEO basics and, later, accessibility and performance. It writes reports for people, CI and coding agents, and runs as a local MCP server so a coding agent can check localhost before it commits.

Use it

npx @webtrax/cli check https://your-site.example            # links, security, SEO
npx @webtrax/cli check https://your-site.example --browser  # + accessibility and performance (needs Chromium)
claude mcp add webtrax -- npx -y @webtrax/cli mcp           # local MCP server for coding agents

GitHub Action: uses: coldbathcode/webtrax-runner/action@v0 with url: (see action/README.md). Cursor (.cursor/mcp.json): { "mcpServers": { "webtrax": { "command": "npx", "args": ["-y", "@webtrax/cli", "mcp"] } } }. The MCP server checks only local targets (localhost, *.local, private IPs) unless started with --allow-public.

Docs: docs.webtrax.dev (getting started, CLI, MCP, CI, every rule).

Related MCP server: maxaeo-ai-visibility-mcp

Develop

Node.js 22.19+ and pnpm 10. pnpm install, then pnpm run check (typecheck, lint, tests; some tests need Chromium: pnpm --filter @webtrax/net-proxy exec playwright-core install chromium) and pnpm run build. See CONTRIBUTING.md.

Package

What

State

@webtrax/protocol

Job spec and run records (Zod + JSON Schema), fingerprints, the WebtraxBot identity

Built (C05.1)

@webtrax/net-guard

Scope policy and guarded fetch: no private networks or cloud metadata, every redirect and DNS answer checked

Built (C05.2)

@webtrax/net-proxy

Egress proxy for the browser stage: Chromium reaches the network only through the scope policy

Built (C05.10)

@webtrax/engine

Crawl engine: one fetch per URL, robots.txt, sitemaps, per-host politeness, 429 handling, time budget and cancellation

Built (C05.3)

@webtrax/suite-links, @webtrax/suite-security, @webtrax/suite-seo

Check suites: broken links, redirects, sitemaps and canonicals; passive security (HTTPS, headers, CSP, mixed content); SEO basics

Built (C05.4)

@webtrax/suite-content

Content: placeholder text left in (lorem ipsum, editor prompts, unrendered template tags, values printed as undefined or NaN), a stale copyright year, pages with almost nothing to read, pages that repeat each other

Built (C06.5, first slice)

@webtrax/suite-ai

AI readiness: llms.txt, AI crawler access in robots.txt (search crawler versus user agent per vendor), structured data that does not parse or is missing; with --browser, pages whose text only appears once JavaScript runs

Built (C06.6, first slice)

@webtrax/reporters

Console, JSON/NDJSON, Markdown (for PRs and LLMs), single-file accessible HTML

Built (C05.5)

@webtrax/cli

webtrax check, init, diff, explain, doctor, and webtrax mcp (local MCP server for coding agents)

Built (C05.6, C05.7)

The crawler identifies itself as Mozilla/5.0 (compatible; WebtraxBot/1.0; +https://webtrax.dev/bot) and honours robots.txt and Crawl-delay. See webtrax.dev/bot.

Licensed under the Apache License 2.0.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      Not graded
      quality
      C
      maintenance
      Enables LLM agents to crawl and audit websites for technical SEO and GEO issues, providing actionable fixes via tools like crawl_site and explain_issue.
      128
      -
    • A
      license
      A
      quality
      C
      maintenance
      Enables AI agents to check whether a public website is crawlable, understandable, and ready for AI search workflows through local-only audits of robots.txt, sitemaps, metadata, and llms.txt.
      3
      79 npm
      2
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables a coding agent to audit any public page the way search engines and AI crawlers see it, including per-crawler robots.txt rules, CDN/firewall blocks, and no-JavaScript content. It returns a prioritized fix pack showing where each fix belongs in the codebase plus a command to verify it, so the agent can apply the changes, flag human-only settings, and re-run the audit to confirm.
      MIT