Labsco
MCP SERVER

Point an audit at a URL and get named findings back — JS errors, 4xx/5xx and slow requests, CORS blocks, security headers, SEO — then Lighthouse scores, four-viewport layout checks and a heap snapshot when the fast pass comes back clean.

Automated Testing & QA✓ Verified
Summary
A QA pass that hands back named findings with severities instead of a waterfall to read.

The audit depths are staged deliberately: argus_audit is the fast detection sweep, and argus_audit_full adds the expensive work — Lighthouse, the four viewports, the heap snapshot — for when the fast pass comes back clean and the bug is still there. Thresholds are fixed rather than advisory, so results are comparable between runs: argus_visual_diff warns at 0.1% of pixels changed and calls 5% critical, and argus_design_audit treats position drift past 20px as a mismatch. One consequence worth planning for: the first argus_visual_diff call on a URL only stores the baseline, so a regression can surface no earlier than the second.

What it is

A Chrome DevTools Protocol QA harness behind 9 tools: two audit depths, screenshot and Figma comparison, environment and pull-request diffs, and live reads of a tab that is already open.

What you get
  • A one-pass detection sweep in argus_audit: JS errors and unhandled rejections, 4xx/5xx network failures, CORS errors, API frequency loops, slow APIs and blocking third-party requests, API contract violations, sync XHR, document.write, long tasks, service worker failures, debugger statements, duplicate IDs, SEO violations and security header checks.
  • A deeper pass in argus_audit_full: Lighthouse performance and accessibility scoring, responsive layout checks at 320, 768, 1280 and 1920px, memory leak detection from a heap snapshot, hover-state regression detection and an accessibility tree snapshot.
  • Live reads of the tab you are already in: argus_watch_snapshot polls console errors, network failures, CORS blocks and auth failures without navigating, and argus_get_context packages everything currently broken — critical issues, warnings, JS errors, network failures, recent requests, open tabs — as a diagnostic to work from.
  • Pixel and design comparison: argus_visual_diff scores a screenshot against a stored baseline with pixelmatch, warning at 0.1% and critical at 5% of pixels changed; argus_design_audit checks a page against a Figma frame across 13 mismatch types, including CSS token values, fill and text colour, typography, Auto Layout padding and gap, border-radius and position drift past 20px.
  • Diffs at the scale you work in: argus_compare runs the analyzer suite on dev and staging and surfaces findings present in staging but not dev, and argus_pr_validate maps a GitHub pull request's changed files to the routes they affect and audits only those.
  • argus_last_report returns the most recent JSON report from the reports/ directory, with its findings array and critical, warning and info counts.
Requirements

Chrome, and a URL the harness can reach. argus_design_audit additionally needs a Figma frame URL, and argus_pr_validate a pull request URL and a GitHub token — its import-graph mapping also wants ARGUS_SOURCE_DIR pointing at the checked-out app source. Reports are written under reports/, screenshot baselines under reports/baselines/screenshots/.

Setup effort

One command — npx -y argusqa-os