BIS Safety — Full-Site Navigation

Purpose

Systematically enumerate and traverse the entirety of bissafety.ca — every marketing page, blog post, and online safety course — and return a structured catalog of URLs (with titles, slugs, taxonomies, and last-modified timestamps) suitable for downstream indexing, link-validation, content-audit, or LLM-ingestion tasks. The site is a WordPress + Yoast SEO marketing/eLearning property; this skill is read-only and exercises the public REST + sitemap surfaces in preference to scripted browsing.

When to Use

• Building a catalog of every BIS Safety course (~1,920 SKUs) for downstream search/comparison.
• Auditing the entire site for broken links, missing canonicals, or content gaps before a migration.
• Bulk-ingesting BIS Safety content (pages, blog posts, course descriptions) into a vector store or knowledge base.
• Generating a navigable site map for an agent that will later deep-link to specific course/page URLs.
• Verifying which URLs the site publicly indexes (e.g., before a robots.txt or noindex audit).
• Snapshotting the site structure on a recurring schedule for change-detection.

Workflow

The site exposes its complete URL inventory through two cheap, structured surfaces — there is no reason to crawl HTML for navigation. Use the REST API + sitemap in tandem; reserve browser navigation for human-visible artifacts (screenshots, rendered hero copy with Elementor blocks, cookie-banner verification).

1. Enumerate URLs via Yoast sitemap index (one HTTP GET → all URLs)

curl -fsSL https://bissafety.ca/sitemap_index.xml

Returns a sitemapindex pointing to four child sitemaps. As of the last verified run:

Each child sitemap returns one <url> per page with <loc>, <lastmod>, and (for pages) inline <image:image> entries. Total enumerable inventory: ~2,238 URLs.

Yoast emits sitemaps with X-Robots-Tag: noindex, follow — they're public but unindexed; you can still fetch them without auth or special headers.

2. Enrich each URL with structured metadata via the WordPress REST API

The /wp-json/wp/v2/ namespace is fully open (no auth required). Custom post type courses is exposed at rest_base courses.

Always use _fields= to whittle the payload — the default response contains content.rendered (full HTML body, ~10–50 KB per item) and yoast_head (a meta-tag dump), which inflate transfer 10× and are rarely needed for navigation tasks.

Pagination contract: pass page=N&per_page=100. Response headers X-WP-Total and X-WP-TotalPages give the totals. Loop until page > X-WP-TotalPages.

The eight course categories (slugs are stable, agent can hardcode for filtering): awareness (700), driver (287), electrical (51), equipment (330), products (6), safety (1373), soft-skills (377), virtual-reality-vr (6). Counts overlap — a course can belong to multiple categories.

Recommended traversal sequence:

1. GET /sitemap_index.xml → list 4 child sitemap URLs.
2. Fetch each child sitemap in parallel → extract <loc> + <lastmod> per URL (regex <loc>([^<]+)</loc> works; full XML parse not required).
3. Bucket URLs by path prefix: /courses/ → courses CPT, /blog/ or /{slug}/ matching a post → posts, all others → pages.
4. For each bucket, page through the matching /wp-json/wp/v2/{rest_base} endpoint with per_page=100 + _fields= to enrich the URL list with id, title.rendered, modified, taxonomy IDs.
5. (Optional) Resolve course-category IDs against /wp-json/wp/v2/course-category once and inline names.

3. Optional: per-URL deep content fetch

If the task needs the rendered prose (e.g., for LLM ingestion):

• Prefer /wp-json/wp/v2/{rest_base}/{id}?_fields=content,excerpt,title,slug,link — clean HTML, no template chrome.
• Fall back to fetching the URL directly with browserless_agent (goto + { "method": "text", "params": { "selector": "body" } }, or parse the DOM in an evaluate). The site serves a Cookie-banner overlay (Complianz/cmplz plugin) on first HTML render — it does not block the underlying DOM, so text extraction works without dismissing the banner.

Browser fallback

If REST is ever disabled, blocked behind the Cloudflare bot check, or you need to verify visually-rendered Elementor content the REST API doesn't expose (some pages use Elementor's _elementor_data postmeta which is not in content.rendered):

1. Call browserless_agent with proxy: { proxy: "residential" } (the site is on Cloudflare + Kinsta + Nitro CDN; residential egress helps, a verified fingerprint is not required). Keep the whole crawl inside one call's commands array so the __cf_bm cookie persists across steps.
2. { "method": "goto", "params": { "url": "https://bissafety.ca", "waitUntil": "load", "timeout": 45000 } } — the homepage's mega-menu exposes every product page in a single render. Site footer + /all-courses/ listing covers the resource hubs.
3. Dismiss the cookie banner only if it overlays a critical hit target — a text/snapshot extraction works without it.
4. Crawl by following anchor hrefs (parse them inside an evaluate), deduplicating by canonical URL. The same-origin filter https://bissafety.ca/ keeps the crawl bounded; explicitly drop cdn-ilegmfm.nitrocdn.com (the Nitro asset CDN — images and JS bundles, not HTML).

Site-Specific Gotchas

• Use the sitemap, not the homepage navigation. The header mega-menu surfaces ~60 product/resource URLs, but the site has ~2,238 total URLs. Skipping the sitemap and following anchors will undercount courses by ~97%.
• Two course sitemaps, not one. Yoast splits sitemaps at 1,000 URLs — both courses-sitemap.xml and courses-sitemap2.xml must be fetched. The sitemap_index lists both; do not stop at the first.
• Old /sitemap.xml 301-redirects to /sitemap_index.xml via Yoast SEO's redirect manager (X-Redirect-By: Yoast SEO). Either URL works, but follow the redirect (curl -L, or a browserless_function same-origin fetch, both honor it).
• WP REST API is fully open — no nonce, no key, no rate limit observed under residential-proxy traffic at ≤5 req/s. CORS is permissive (Access-Control-Allow-Origin not restricted), so the agent does not need to spoof Origin headers.
• per_page max is 100. Asking for more (e.g., per_page=500) returns a 400 rest_invalid_param. Use page=N to paginate.
• Always pass _fields=. Default REST payload includes the full rendered HTML body, yoast_head (~5 KB of meta tags per item), and _links (~2 KB HAL). Whittling to id,slug,link,title,date,modified,course-category cuts 1,920-course enumeration from ~80 MB to ~3 MB.
• Course CPT taxonomy field is course-category (hyphen, not underscore). The REST _fields selector and the filter parameter both use the hyphenated form: ?course-category=42 to filter by term ID, or ?course-category=safety does not work — you must resolve the slug to an ID first via /wp-json/wp/v2/course-category?slug=safety.
• Slug collisions across post types. Several pages have slug=lp under different parents (/company-spotlights/lp/, /ai-in-the-workplace/lp/). Always key on link (full URL) or id, not slug, when deduping.
• __cf_bm cookie is set by Cloudflare on every response. Reusing a session that holds it materially speeds up subsequent fetches (drops latency from ~600 ms to ~120 ms by skipping bot-check); keeping the whole flow inside one browserless_agent (or browserless_function) call persists cookies automatically across its steps.
• Nitro CDN cache (X-Nitro-Cache: HIT) front-runs Kinsta. Pages refreshed minutes ago may still serve a stale <lastmod> in the sitemap until the Nitro purge fires — for monotonically fresh data, prefer the REST API's modified field over the sitemap's <lastmod>.
• Cookie-consent banner (Complianz / cmplz) overlays the page on the first HTML render but does not block underlying DOM access. a text/html extraction (or a snapshot) both see through it. There is no need to click "Accept" to extract content.
• No GraphQL endpoint. /graphql returns 404 — do not waste cycles on WPGraphQL-style queries.
• Image CDN domain is different. All assets serve from cdn-ilegmfm.nitrocdn.com (Nitro Pack). If you build a media inventory, dedupe by the original bissafety.ca/wp-content/uploads/... path that the CDN URL wraps.
• Author archives, search, and feeds are disallowed in robots.txt (/author/, /search/, /feed/, /?s=) — respect that boundary; they're explicitly excluded from "the entire site" surface for this skill.
• Some Elementor pages embed content via _elementor_data postmeta which the REST API does not include in content.rendered. Twelve to twenty pages (notably product landing pages built in Elementor Pro) will look near-empty via REST; for those, the browser fallback's text/markdown extraction is the source of truth.
• /wp-json/wp/v2/users returns 401 rest_user_cannot_view for unauthenticated callers (good — author enumeration is blocked). Don't try to map post authors without credentials.
• Recurring slug pattern *-course-subscription and *-sitemap*.xml are the canonical anchor points for the subscription bundles and the sitemap surface, respectively. Useful for regex-bucketing.

Expected Output

A JSON document with one top-level urls array plus per-bucket counts and a discovery manifest. The shape an agent SHOULD produce:

{
  "domain": "bissafety.ca",
  "discovered_at": "2026-05-25T23:13:00Z",
  "source": "sitemap_index+wp_rest_api",
  "counts": {
    "pages": 98,
    "posts": 220,
    "courses": 1920,
    "total": 2238
  },
  "course_categories": [
    { "slug": "safety", "name": "Safety", "count": 1373 },
    { "slug": "awareness", "name": "Awareness", "count": 700 },
    { "slug": "soft-skills", "name": "Soft Skills", "count": 377 },
    { "slug": "equipment", "name": "Equipment", "count": 330 },
    { "slug": "driver", "name": "Driver", "count": 287 },
    { "slug": "electrical", "name": "Electrical", "count": 51 },
    { "slug": "products", "name": "Products", "count": 6 },
    { "slug": "virtual-reality-vr", "name": "Virtual Reality (VR)", "count": 6 }
  ],
  "urls": [
    {
      "type": "page",
      "id": 62038,
      "slug": "homepage",
      "link": "https://bissafety.ca/",
      "title": "EHS Software & Safety Management Platform | BIS Software",
      "modified": "2026-05-15T14:46:37Z"
    },
    {
      "type": "post",
      "id": 63921,
      "slug": "safety-spotlight-building-real-safety-culture-erin-heimbecker",
      "link": "https://bissafety.ca/safety-spotlight-building-real-safety-culture-erin-heimbecker/",
      "title": "Saskatchewan Association for Safe Workplaces in Health (SASWH) – From the Field to the Floor: Building Real Safety Culture with Erin Heimbecker",
      "modified": "2026-05-06T12:50:19Z",
      "categories": [12, 47]
    },
    {
      "type": "course",
      "id": 64104,
      "slug": "active-shooter-active-threat-organizational-preparedness-recovery",
      "link": "https://bissafety.ca/courses/active-shooter-active-threat-organizational-preparedness-recovery/",
      "title": "Active Shooter/Active Threat: Organizational Preparedness & Recovery",
      "modified": "2026-05-21T08:07:24Z",
      "course-category": ["safety", "awareness"]
    }
  ]
}

If the task asks for navigation in a tree shape (mega-menu top-level sections) rather than a flat URL list, the alternative shape:

{
  "domain": "bissafety.ca",
  "discovered_at": "2026-05-25T23:13:00Z",
  "navigation": {
    "software": {
      "ehs_platform": [
        {
          "title": "Health and Safety Software",
          "link": "https://bissafety.ca/health-and-safety-software/"
        },
        {
          "title": "Safety Management System (SMS)",
          "link": "https://bissafety.ca/safety-management-system-sms/"
        },
        {
          "title": "Learning Management System (LMS)",
          "link": "https://bissafety.ca/learning-management-system-lms/"
        }
      ],
      "safety_training": [
        {
          "title": "Online Orientation Software",
          "link": "https://bissafety.ca/online-orientation-software/"
        },
        {
          "title": "Virtual Proctoring",
          "link": "https://bissafety.ca/virtual-proctoring/"
        }
      ],
      "industry": [
        {
          "title": "Transportation",
          "link": "https://bissafety.ca/ehs-software-for-the-transportation-industry/"
        },
        {
          "title": "Energy",
          "link": "https://bissafety.ca/ehs-software-for-the-energy-industry/"
        },
        {
          "title": "Construction",
          "link": "https://bissafety.ca/construction-industry-ehs-software/"
        }
      ]
    },
    "courses": {
      "all_courses_landing": "https://bissafety.ca/all-courses/",
      "subscriptions": "https://bissafety.ca/course-subscription-plans/",
      "total": 1920
    },
    "resources": {
      "blog": "https://bissafety.ca/blog/",
      "podcasts": "https://bissafety.ca/safety-spotlight-podcasts/",
      "events": "https://bissafety.ca/events/",
      "magazine": "https://bissafety.ca/safetynet-magazine/",
      "company_spotlights": "https://bissafety.ca/company-spotlights/"
    },
    "company": {
      "about": "https://bissafety.ca/about-us/",
      "careers": "https://bissafety.ca/careers/",
      "testimonials": "https://bissafety.ca/testimonial/",
      "faq": "https://bissafety.ca/frequently-asked-questions/",
      "contact": "https://bissafety.ca/contact-us/",
      "demo": "https://bissafety.ca/request-a-demo/",
      "legal_trust_centre": "https://bissafety.ca/legal-trust-centre/"
    }
  }
}

If the REST API is unreachable and only the sitemap was harvested, return the sitemap shape (still a valid full-site enumeration, just lacking taxonomy/title enrichment):

{
  "domain": "bissafety.ca",
  "discovered_at": "2026-05-25T23:13:00Z",
  "source": "sitemap_only",
  "counts": { "pages": 98, "posts": 220, "courses": 1913, "total": 2231 },
  "urls": [
    {
      "type": "page",
      "link": "https://bissafety.ca/",
      "lastmod": "2026-05-15T14:46:37Z"
    },
    {
      "type": "post",
      "link": "https://bissafety.ca/transportation-safety-week/",
      "lastmod": "2026-05-22T23:35:45Z"
    },
    {
      "type": "course",
      "link": "https://bissafety.ca/courses/whmis-2025/",
      "lastmod": "2026-05-21T08:07:24Z"
    }
  ]
}