# ============================================================
# adamkakembo.com — robots.txt
# Last updated: 2026-05-28 — add ai-usage-terms, CCBot, Claude-User
#
# Policy:
#   Legitimate search and AI discovery crawlers: ALLOWED
#   Bulk scrapers, content harvesters, cloners: DISCOURAGED
#   Private paths, backups, dev files: BLOCKED
# ============================================================


# ──────────────────────────────────────────────────────────
# Global: allow all legitimate crawlers by default
# ──────────────────────────────────────────────────────────
User-agent: *
Allow: /
Disallow: /archive/
Disallow: /audio-player-test.html
Disallow: /audio-preview.html
Disallow: /email-signature.html
Disallow: /index.backup-2026-05-25.html
Disallow: /styles.backup-2026-05-25.css
Disallow: /main.backup-2026-05-25.js
Disallow: /operating-profile-PREVIEW.html
Disallow: /audio/
Crawl-delay: 10

Sitemap: https://adamkakembo.com/sitemap.xml

# AI guidance: https://adamkakembo.com/llms.txt


# ──────────────────────────────────────────────────────────
# Major search engines — full access, no delay
# ──────────────────────────────────────────────────────────
User-agent: Googlebot
Allow: /
Disallow: /archive/
Disallow: /audio-player-test.html
Disallow: /audio-preview.html
Disallow: /email-signature.html
Disallow: /index.backup-2026-05-25.html
Disallow: /styles.backup-2026-05-25.css
Disallow: /main.backup-2026-05-25.js
Disallow: /operating-profile-PREVIEW.html

User-agent: Googlebot-Image
Allow: /images/

User-agent: Bingbot
Allow: /
Disallow: /archive/
Disallow: /audio-player-test.html
Disallow: /audio-preview.html
Disallow: /email-signature.html
Disallow: /index.backup-2026-05-25.html
Disallow: /styles.backup-2026-05-25.css
Disallow: /main.backup-2026-05-25.js
Disallow: /operating-profile-PREVIEW.html

User-agent: DuckDuckBot
Allow: /
Disallow: /archive/

User-agent: Slurp
Allow: /
Disallow: /archive/

User-agent: Baiduspider
Allow: /
Disallow: /archive/

User-agent: Applebot
Allow: /
Disallow: /archive/


# ──────────────────────────────────────────────────────────
# Legitimate AI crawlers — discovery and citation allowed
# Permitted: read, summarise, cite, attribute, reference
# Prohibited: bulk extraction, wholesale reproduction,
#             training data collection without permission
# ──────────────────────────────────────────────────────────
User-agent: GPTBot
Allow: /
Disallow: /archive/
Crawl-delay: 15

User-agent: ChatGPT-User
Allow: /
Disallow: /archive/
Crawl-delay: 15

User-agent: Claude-Web
Allow: /
Disallow: /archive/
Crawl-delay: 15

User-agent: ClaudeBot
Allow: /
Disallow: /archive/
Crawl-delay: 15

User-agent: PerplexityBot
Allow: /
Disallow: /archive/
Crawl-delay: 15

User-agent: Google-Extended
Allow: /
Disallow: /archive/
Crawl-delay: 10

User-agent: Gemini
Allow: /
Disallow: /archive/
Crawl-delay: 10

User-agent: Meta-ExternalAgent
Allow: /
Disallow: /archive/
Crawl-delay: 15

User-agent: Meta-ExternalFetcher
Allow: /
Disallow: /archive/
Crawl-delay: 15

User-agent: Applebot-Extended
Allow: /
Disallow: /archive/
Crawl-delay: 10

User-agent: cohere-ai
Allow: /
Disallow: /archive/
Crawl-delay: 15

User-agent: anthropic-ai
Allow: /
Disallow: /archive/
Crawl-delay: 15

User-agent: YouBot
Allow: /
Disallow: /archive/
Crawl-delay: 15

User-agent: Diffbot
Allow: /
Disallow: /archive/
Crawl-delay: 15

User-agent: Claude-User
Allow: /
Disallow: /archive/
Crawl-delay: 15

User-agent: CCBot
Allow: /
Disallow: /archive/
Crawl-delay: 15


# ──────────────────────────────────────────────────────────
# Aggressive scrapers and known content harvesters — blocked
# These bots are associated with bulk crawling, content
# cloning, dataset extraction or competitive intelligence
# scraping unrelated to legitimate search or AI citation.
# ──────────────────────────────────────────────────────────
User-agent: AhrefsBot
Disallow: /

User-agent: SemrushBot
Disallow: /

User-agent: MJ12bot
Disallow: /

User-agent: DotBot
Disallow: /

User-agent: BLEXBot
Disallow: /

User-agent: DataForSeoBot
Disallow: /

User-agent: PetalBot
Disallow: /

User-agent: serpstatbot
Disallow: /

User-agent: SEOkicks
Disallow: /

User-agent: linkdexbot
Disallow: /

User-agent: Screaming Frog SEO Spider
Disallow: /

User-agent: rogerbot
Disallow: /

User-agent: spbot
Disallow: /

User-agent: Exabot
Disallow: /

User-agent: ia_archiver
Disallow: /archive/

User-agent: wget
Disallow: /

User-agent: curl
Disallow: /