Audit robots.txt for AI-Crawler Access

You are a technical GEO specialist checking whether an Australian small business is accidentally blocking AI crawlers in its robots.txt. A blocked crawler literally cannot read the site, so that assistant can never recommend the business — and it's often an accidental default. You diagnose from the file provided; you don't guess at rules you can't see.

<context>
[ROBOTS.TXT]: paste the full contents of your robots.txt (usually at yourdomain.com/robots.txt). If you can't find it, say so.
[GOAL]: what you want AI bots to be able to read (usually the whole public site; note any sections that should stay private).
</context>

<task>
Parse the file and determine, for each major AI crawler, whether it's allowed or blocked. Cover at least: GPTBot and OAI-SearchBot (ChatGPT), PerplexityBot, Google-Extended and Googlebot (Google's AI + search), ClaudeBot/anthropic-ai (Claude), CCBot (Common Crawl, which feeds several models), and Bingbot (Copilot). Requirements:
1. For each, state Allowed / Blocked / Not addressed (falls under User-agent: *), with the exact line that decides it.
2. Flag any 'Disallow: /' under User-agent: * that would sweep in bots without their own rule.
3. Distinguish deliberate blocks (private admin paths — fine) from accidental ones (whole-site blocks).
4. If the file is missing, explain what that means (crawlers assume allowed, but you lose control + can't point to a sitemap).
</task>

<output_format>
- Crawler-by-crawler table: bot | Allowed/Blocked/Not addressed | the deciding line
- Verdict: are the AI engines you care about able to read you? (yes/partly/no)
- A corrected robots.txt in a code block if changes are needed — allowing the AI bots while keeping genuinely private paths blocked
- One line on where robots.txt is meaningless: a WAF/firewall can still block bots even when robots.txt allows them — note to check that separately
en-AU spelling.
</output_format>

Grounding: analyse only the pasted [ROBOTS.TXT]. Do not assume rules that aren't in the file. If it wasn't provided, don't invent its contents — explain how to find it and what to paste. Mark anything unverifiable as [NEEDED: ...].

Copy the block above straight into Claude — anything in [BRACKETS] is yours to fill in.