site-chatbot-wiki

name

site-chatbot-wiki

description

Generate a website-grounded wiki context bundle for chatbot use. Use this skill when the user wants to turn a public website into structured markdown docs for retrieval and answering visitor questions in a website chatbot. Trigger on phrases like '/site-chatbot-wiki', 'website chatbot context', 'build chatbot knowledge base from website', 'generate docs from website URL', 'create a website FAQ wiki', and 'chatbot resource from my website'.

site-chatbot-wiki

Build a dense, LLM-readable wiki from website content so chatbots can answer visitor questions with high precision and low hallucination. The output is optimized for retrieval and context injection.

What this produces

A wiki/ directory containing:

• _index.md — table of contents with one-line summaries
• _context.md — concatenated context bundle for LLM prompt injection
• [topic].md — one article per chatbot-relevant topic
• AGENTS.md (optional) — IDE and assistant context-loading helper

Best-fit use cases

• Marketing and brochure websites
• Service business sites (consulting, healthcare, childcare, legal, local services)
• Product landing sites with pricing, policies, and FAQ
• Multi-page websites where answers must stay aligned to published content

Not ideal without extra inputs

• Highly dynamic SaaS dashboards hidden behind auth
• Websites with critical facts only in PDFs/images unless content is provided
• Domains requiring legal compliance interpretation beyond published copy

Article template

Every generated article must follow this structure:

# [Title]

> **TL;DR**: One-sentence summary of what this article covers.

## Overview
2-4 sentences defining the topic and why visitors ask about it.

## [Key Section 1]
...

## [Key Section 2]
...

## Visitor Questions This Covers
- Q: [Common user question]
  A: [Grounded answer from website context]

## Examples
Concrete examples that mirror real visitor questions.

## Related
- [[other-article-slug]] - one-line description
- [[another-article-slug]] - one-line description

## Overview and ## Related are mandatory for every article.

Workflow

Step 1 - Parse the prompt

Extract:

• Website URL(s)
• Audience (site visitors, sales leads, existing customers, parents, etc.)
• Desired chatbot scope (support, sales, FAQ, intake pre-qualification)
• Output location (default: ./wiki/)

If critical information is missing, ask one clarifying question.

Step 2 - Plan the article set

Before writing articles, output:

## Wiki Plan: [Website/Brand]

Articles to generate (N total):
1. `slug-name` - [Title] - [one-line scope]
2. ...

Audience: [who]
Depth: [shallow overview / standard / deep]
Output: [path]

Target 8-15 articles for most websites.

Step 3 - Generate chatbot-ready articles

Prioritize website-chatbot topics:

• brand-overview
• services
• pricing-and-packages
• service-areas or locations
• booking-or-contact-flow
• faq
• policies (refunds, cancellation, privacy, terms)
• trust-signals (testimonials, certifications, case studies)

For each article:

1. Create wiki/[slug].md
2. Keep 150-450 lines unless the topic requires splitting
3. Include concrete visitor-style examples and Q/A snippets

Step 4 - Generate _index.md and _context.md

wiki/_index.md must contain:

• Wiki title and short scope paragraph
• Article table with summary per slug
• Usage notes for humans and AI assistants

wiki/_context.md must concatenate all articles with separators:

<!-- site-chatbot-wiki context bundle | domain: [domain] | articles: N -->

---
# [Article 1 title]
[full content]

Step 5 - Add chatbot guardrail file (recommended)

Create wiki/chatbot-guardrails.md with:

• Allowed answer scope (public website facts only)
• Unknown-answer behavior
• Escalation triggers (contact form, phone, email)
• Disallowed content and compliance disclaimers

Quality checklist

Before finishing:

• Every article has ## Overview and ## Related
• Every slug referenced in ## Related exists
• _index.md matches generated files
• _context.md contains all articles
• Content is website-grounded and avoids invented details
• Visitor questions are covered with direct, concise answers

Example invocations

• /site-chatbot-wiki build reference docs for my website https://bestitconsulting.ca as a chatbot resource
• /site-chatbot-wiki generate context for my website https://bestitconsultants.ca so the wiki can be used as its website chatbot
• /site-chatbot-wiki create a knowledge base for website https://friendshipdaycare.ca as a website chatbot to answer visitors questions

Tips

• Keep answers grounded in pages that actually exist.
• Prefer explicit numbers, policies, and process steps over vague summaries.
• If website data is incomplete, record assumptions clearly in the article.