| name | safari-seo |
| description | Open-source agentic SEO skill for Tanzania safari, tour-operator, and travel sites. Works in Grok, Cursor, Claude Code, Codex, Copilot. Triggers on: safari SEO, tour operator SEO, Kilimanjaro/Serengeti/Ngorongoro ranking, TouristTrip schema, migration content, Northern Circuit keywords, local SEO Arusha Moshi, safari audit, hreflang, Core Web Vitals wildlife photos, or "perform SEO analysis on <url>". Ships 363 keywords, 15 sub-skills, 14 scripts. Outputs FULL-AUDIT-REPORT.md + ACTION-PLAN.md. Repo: github.com/cleven12/tanzania-tourism-seo
|
Safari SEO Skill (Agentic / Tanzania Tourism)
LLM-first SEO analysis skill purpose-built for Tanzania safari and tour-operator websites. Includes 15 specialized sub-skills, 7 specialist agents, and script-backed evidence collection for Kilimanjaro treks, Serengeti safaris, Ngorongoro crater trips, Northern Circuit packages, and Zanzibar add-ons.
<SKILL_DIR> = absolute path to this skill directory (the folder containing this SKILL.md).
Deterministic Trigger Mapping
Map common user wording to a fixed workflow:
| User says | Route to |
|---|
perform seo analysis on <url> | Single-URL full audit → resources/skills/safari-page.md |
audit my safari site / full SEO check | Full site audit → resources/skills/safari-audit.md |
optimize tour page / new safari package | resources/skills/safari-tour-page.md |
serengeti page / kilimanjaro landing page | resources/skills/safari-destination.md |
keyword research tanzania | resources/skills/safari-keywords.md |
google business profile arusha | resources/skills/safari-local.md |
schema for safari package | resources/skills/safari-schema.md |
hreflang swahili english | resources/skills/safari-hreflang.md |
competitor analysis safari | resources/skills/safari-competitors.md |
AI search / ChatGPT citations | resources/skills/safari-geo.md |
| No explicit sub-skill + URL provided | Single-URL full audit with mandatory deliverables |
For full/page audits, always produce:
FULL-AUDIT-REPORT.md (detailed findings with evidence)ACTION-PLAN.md (prioritized fixes)
Available Commands
| Command | Sub-Skill | Description |
|---|
safari audit <url> | safari-audit | Full site audit with Tanzania scoring |
safari page <url> | safari-page | Deep single-page analysis |
safari content <url> | safari-content | E-E-A-T, depth, safari trust signals |
safari technical <url> | safari-technical | Crawlability, CWV, indexing, security |
safari schema <url> | safari-schema | TouristTrip, Product, TouristAttraction JSON-LD |
safari local <url> | safari-local | GBP, citations, Arusha/Moshi local SEO |
safari keywords <topic> | safari-keywords | Park-level keyword & seasonality research |
safari plan <url> | safari-plan | Strategic content calendar & hub architecture |
safari competitors <query> | safari-competitors | SERP gap analysis vs ranking operators |
safari hreflang <url> | safari-hreflang | EN/DE/FR/SW international validation |
safari images <url> | safari-images |
Orchestration Logic
Step 1 — Identify the Task
Parse the user's request and read the matching resources/skills/safari-*.md file before executing.
- Full audit:
safari-audit.md — crawl key pages (homepage, top tours, destination hubs, blog), delegate to agents
- Single page:
safari-page.md — deep dive on one URL
- Tour launch:
safari-tour-page.md + checklists/new-tour-page.md
- Destination hub:
safari-destination.md + checklists/new-destination-page.md
- Strategic plan:
safari-plan.md + seasonality from 04-tanzania-niche/seasonality.md
Step 2 — Load Tanzania Ground Truth
Always load references.md when citing fees, visas, park rules, or authority facts.
| Fact type | Authoritative source | Never |
|---|
| National park fees | tanzaniaparks.go.tz (TANAPA) | Guess USD amounts |
| Ngorongoro fees | ncaa.go.tz | Conflate with TANAPA |
| Visa rules | immigration.go.tz | Outdated blog posts |
| Industry standards | tatotz.org (TATO) | Invent certifications |
| Porter ethics | kpap.org | Generic "responsible tourism" |
| Keyword volumes | 04-tanzania-niche/keywords/*.csv bands | Fabricate exact MSV |
Step 3 — Collect Evidence
Primary (LLM-first): Use read_url_content or direct page fetch for baseline reasoning.
Deterministic verification (when script execution is available):
python3 <SKILL_DIR>/scripts/fetch_page.py <url> --output /tmp/safari-page.html
python3 <SKILL_DIR>/scripts/parse_html.py /tmp/safari-page.html --url <url> --json
python3 <SKILL_DIR>/scripts/seo-audit.py <url> --json
python3 <SKILL_DIR>/scripts/pagespeed.py <url> --strategy mobile --json
python3 <SKILL_DIR>/scripts/robots_checker.py <url>
python3 <SKILL_DIR>/scripts/security_headers.py <url>
python3 <SKILL_DIR>/scripts/redirect_checker.py <url>
python3 <SKILL_DIR>/scripts/validate_schema.py <file_or_url>
python3 <SKILL_DIR>/scripts/generate-schema.py --template touristtrip --input tour.yaml --json
python3 <SKILL_DIR>/scripts/sitemap_checker.py <url>
python3 <SKILL_DIR>/scripts/sitemap-gen.py --urls urls.txt --output sitemap.xml
python3 <SKILL_DIR>/scripts/image_inventory.py <url> --json
python3 <SKILL_DIR>/scripts/image_weight_audit.py <url> --json
python3 <SKILL_DIR>/scripts/local_seo_checker.py <url> --json
python3 <SKILL_DIR>/scripts/hreflang_checker.py <url> --json
python3 <SKILL_DIR>/scripts/readability.py /tmp/safari-page.html --json
python3 <SKILL_DIR>/scripts/article_seo.py <url> --keyword "<target_keyword>" --json
python3 <SKILL_DIR>/scripts/broken_links.py <url> --workers 5
python3 <SKILL_DIR>/scripts/llms_txt_checker.py <url>
python3 <SKILL_DIR>/scripts/citation_readiness.py <url> --json
Do not use third-party mirrors (e.g., r.jina.ai) as primary evidence when direct fetch or bundled scripts are available.
If a script fails (network, timeout, missing dependency), report as environment limitation — not a confirmed site issue. Continue with LLM analysis at Likely confidence.
Step 4 — Delegate to Specialist Agents
For comprehensive audits, adopt the relevant agent role from resources/agents/:
| Agent | File | Focus |
|---|
| Technical SEO | safari-technical.md | Crawlability, CWV, mobile, JS rendering |
| Content Quality | safari-content.md | E-E-A-T, experience signals, thin content |
| Schema Markup | safari-schema.md | TouristTrip, Product, TouristAttraction |
| Local SEO | safari-local.md | GBP, NAP, Arusha/Moshi citations |
| Performance | safari-performance.md | LCP heroes, INP on booking widgets |
| Verifier | safari-verifier.md | Deduplicate, validate evidence, suppress contradictions |
| Tanzania Niche | tanzania-niche.md | Migration timing, routes, fees, seasonality |
Step 5 — Apply Repo Quality Gates
| Resource | Path | Use for |
|---|
| Meta tag formulas | 01-onpage-content/meta-tags-guide.md | Title/description patterns |
| Page architecture | 01-onpage-content/page-architecture.md | Hub-and-spoke tour/destination structure |
| Internal linking | 01-onpage-content/internal-linking.md | Blog → tour → destination flows |
| Content briefs | 01-onpage-content/content-briefs/ | Kilimanjaro, Serengeti, Ngorongoro templates |
| Schema templates | 02-technical/schema/ | JSON-LD with {{PLACEHOLDER}} syntax |
| Crawl checklist | 02-technical/crawl-checklist.md | Indexing gates |
| CWV thresholds | 02-technical/core-web-vitals.md | LCP/INP/CLS targets |
| Hreflang guide | 02-technical/hreflang-i18n.md | EN/DE/FR/SW patterns |
| GBP guide | 03-local/google-business-profile.md | Operator categories, service areas |
| Seasonality | 04-tanzania-niche/seasonality.md | Content calendar timing |
| Search intent | 04-tanzania-niche/search-intent-by-park.md | Info vs commercial routing |
| Glossary | 04-tanzania-niche/glossary.md | Swahili terms, TANAPA/TATO/KPAP |
| Competitor patterns | 04-tanzania-niche/competitor-patterns.md | SERP observation framework |
| Launch checklists | checklists/ | Tour page, destination, monthly SEO |
Step 6 — Verify Findings
python3 <SKILL_DIR>/scripts/finding_verifier.py --findings-json <raw_findings.json> --json
Use verified output for final report tables. Cross-check Tanzania-specific claims against references.md authorities.
Step 7 — Score and Report
Tanzania Safari Scoring Weights (Full Audit)
Canonical source of truth — defined here and mirrored in resources/skills/safari-audit.md.
| Category | Weight | Rationale |
|---|
| Content Quality & E-E-A-T | 22% | Trust-critical YMYL-adjacent travel queries |
| Technical SEO | 18% | Mobile booking, crawlability, indexation |
| On-Page SEO | 15% | Title/meta/H1 for competitive safari SERPs |
| Schema / Structured Data | 15% | TouristTrip, Product, Offer rich results |
| Local SEO & Trust | 12% | Arusha/Moshi operators depend on local + GBP |
| Performance (CWV) | 8% | Hero images and booking widgets affect CWV |
| Image Optimization | 5% | Wildlife photos are primary content |
| AI Search Readiness (GEO) | 5% | Growing AI travel research channel |
Score Interpretation
| Score | Rating |
|---|
| 90-100 | Excellent |
| 70-89 | Good |
| 50-69 | Needs Improvement |
| 30-49 | Poor |
| 0-29 | Critical |
Step 8 — Mandatory Deliverables
For safari audit, safari page, and generic perform seo analysis on <url> flows:
- Create
FULL-AUDIT-REPORT.md at audit start; update as evidence arrives
- Create
ACTION-PLAN.md at audit start; prioritize Critical → High → Medium → Low
- List all generated artifact paths in the final response
- Include Environment Limitations section if scripts were blocked
- Flag any unverified fee/visa claims with
⚠️ VERIFY AT SOURCE
Report Severity Levels
- 🔴 Critical — Blocks indexing, violates structured data policy, or misstates fees/visas
- ⚠️ Warning — Optimization opportunity (fix within 1 month)
- ✅ Pass — Meets or exceeds standards
- ℹ️ Info — Not applicable or informational
Tanzania Page-Type Detection
Auto-detect page type to apply correct checks:
| Signals | Page Type | Sub-Skill | Min Words |
|---|
| Price, duration, itinerary, "book now" | Tour package | safari-tour-page.md | 800+ |
| Park name, wildlife, best time, no price | Destination hub | safari-destination.md | 1,200+ |
| Route comparison, packing list, how-to | Informational blog | safari-content.md | 1,500+ |
| Company name, team, licenses, contact | Operator brand | safari-local.md | 500+ |
/de/, /fr/, hreflang | International | safari-hreflang.md | — |
Schema Templates (Repo)
Pre-built JSON-LD in 02-technical/schema/:
| Template | Page Type |
|---|
touristtrip.json + product-offer-tour.json | Multi-day safari/trek packages |
touristattraction.json | Park/destination hubs |
organization-travelagency.json | Operator footer sitewide |
review-aggregaterating.json | First-party reviews only |
breadcrumblist.json | All pages with URL depth |
faqpage.json | FAQ sections (note: rich results restricted for commercial FAQ schema — use for content structure, not rich result expectation) |
Generate programmatically:
python3 <SKILL_DIR>/scripts/generate-schema.py --template product-offer-tour --input tour.yaml
Validate before publish:
python3 <SKILL_DIR>/scripts/validate_schema.py templates/tour-page.html
Critical Rules (Tanzania Safari)
- INP not FID — INP is the sole interactivity metric (FID removed September 9, 2024). Never reference FID.
- Never fabricate fees — Cite
tanzaniaparks.go.tz or ncaa.go.tz with retrieval date; use ⚠️ VERIFY if live fetch failed.
- Never fabricate reviews in schema —
AggregateRating must reflect real first-party reviews only.
- FAQ schema is restricted — FAQPage rich results limited to government/health authority sites. Commercial safari sites may use FAQ content structure without expecting FAQ rich results.
- HowTo schema is deprecated — Never recommend for Kilimanjaro packing lists.
- JSON-LD only — Never recommend Microdata or RDFa.
- Migration claims need dates — Serengeti migration timing varies annually; cite general patterns from
04-tanzania-niche/seasonality.md, avoid false precision ("crossing on July 15").
- Kilimanjaro route facts — Seven official routes (Marangu, Machame, Lemosho, Shira, Rongai, Northern Circuit, Umbwe); success rates vary by operator — don't invent percentages without source.
- White-hat only — Refuse PBNs, fake reviews, scraped aggregator content, AI doorway pages per park.
- Location page limits — Warning at 30+ near-duplicate destination pages; hard stop at 50+.
- LLM-first, script-backed — Always attempt scripts; never block report on single script failure.
- Mandatory file artifacts —
FULL-AUDIT-REPORT.md + ACTION-PLAN.md for audit flows.
- Commercial vs informational routing — Route info queries to blogs (
safari-content); commercial queries to tour pages (safari-tour-page) per 04-tanzania-niche/search-intent-by-park.md.
Dependencies
pip install -r <SKILL_DIR>/scripts/requirements.txt
pip install playwright && playwright install chromium
Local repo scripts (M4+): seo-audit.py, sitemap-gen.py, generate-schema.py — see scripts/README.md.
