Overview

HK-101 Living RAG provides answers by retrieving and synthesizing information from local text or Markdown files using a retrieval-augmented generation approach. It operates against a local docsPath (default ./docs relative to CWD) with a top-k (default 3) to tailor results, enabling privacy-preserving Q&A from your own docs.

How This Skill Works

The tool searches the local docsPath for the query, retrieves the top-k matches, and feeds the snippets to an LLM to synthesize a concise answer. The output includes an answer and a list of matches with path, score, and snippet. It requires OPENAI_API_KEY to be set in the environment.

When to Use It

• You have internal manuals or knowledge bases stored as Markdown or text in a local folder.
• You need to answer questions while keeping data on-premises to protect privacy.
• You want the LLM to be limited to information from your own docs (no external web data).
• You require quick Q&A for product or policy documentation without building a separate index.
• You need to provide synthesized answers with source snippets for audits or compliance.

Quick Start

1. Step 1: Place your Markdown or text docs under ./docs (or your preferred docsPath).
2. Step 2: Set your inputs: query, optional docsPath, and k (default 3); ensure OPENAI_API_KEY is in the environment.
3. Step 3: Run the query to receive an synthesized answer and a list of matching snippets with their sources.

Best Practices

• Organize docs in ./docs with clear structure and consistent headings to improve matching.
• Start with a small k (e.g., 3) and adjust up or down based on result quality and latency.
• Keep the docs focused on high-value questions to reduce noise in matches.
• Ensure OPENAI_API_KEY is securely set in the environment and monitor usage/costs.
• Validate snippets in matches to verify answer accuracy before user-facing deployment.

Example Use Cases

• Answering internal product manual questions from the local docs folder.
• Retrieving API usage details and examples from codebase markdown files.
• Summarizing quarterly reports stored as local documentation for quick briefs.
• Generating customer-support FAQs from a knowledge base of PDFs/markdowns.
• Onboarding new employees with policies and procedures drawn from internal docs.

Frequently Asked Questions