Personal Rag Builder
Build a private, local RAG system to index and search your files and repos without using third-party APIs.
$9
Works with the AI Agents you already use
personal-rag-builder
Example session with this skill installed
Build a local RAG index for my 'projects/docs' folder using BM25 for keyword search.
- Read your context and instructions
- Compiled the personal-rag-builder
I have initialized a local RAG index for your documents. The system has processed 42 files into 156 chunks and created a BM25 keyword index stored in your local directory.
Local RAG Index Summary
| Component | Status | Details |
|---|---|---|
| Source Path | Active | /projects/docs |
| Total Chunks | Completed | 156 chunks (approx 300 words each) |
| Index Type | Active | BM25 Keyword Search |
| Privacy Mode | Verified | 100% Local (No external API calls) |
| Search Script | Generated | search_index.py |
import json
from rank_bm25 import BM25Okapi
def search(query, top_n=3):
# Load local index and chunks
with open('index_store.json', 'r') as f:
data = json.load(f)
tokenized_query = query.lower().split()
bm25 = BM25Okapi([c['tokens'] for c in data['chunks']])
results = bm25.get_top_n(tokenized_query, data['chunks'], n=top_n)
return [{"text": r['text'], "source": r['source']} for r in results]
# Example usage: print(search("authentication logic"))
Next steps
- Run
python search_index.pyto test retrieval accuracy. - Add your
.envor.gitfolders to the exclusion list if they contain sensitive secrets. - Schedule a refresh task to update the index when you modify your documentation files.
Connects securely to your tools. The creator never sees your data.
About this skill
Building a RAG system often involves complex cloud infrastructure and third-party APIs that compromise data privacy. This skill provides a blueprint for creating a private retrieval index over local files, notes, and repositories without sending sensitive data to external servers.
What it does
- Local indexing creates a searchable database of user-owned files, notes, and public repositories.
- Privacy-first architecture ensures all chunks and indices stay on the local device.
- Hybrid search builds a BM25 keyword index for zero-cost retrieval with optional vector embeddings.
- Source tracking attaches specific file origins to every chunk for verifiable citations.
- Index maintenance handles refreshing data when files change or removing specific chunks.
How it works
- Source collection gathers the specific workspace files or repositories the user wants to index.
- Chunking splits documents into small, manageable segments with metadata labels.
- Index building generates the keyword and vector stores locally.
- Search implementation adds a script to query the index and return the most relevant chunks.
Frameworks & tools
This skill works with local file systems and Python-based indexing libraries. It uses the BM25 algorithm for keyword matching and supports small local embedding models for vector search.
Why this beats prompting it yourself
Prompting an agent to "remember" files often leads to hallucinations or context window exhaustion. This skill provides a systematic engineering approach to data retrieval that maintains a verifiable link between questions and local source material.
Use cases
- Search across personal engineering notes and technical documentation.
- Query local code repositories to find specific logic or patterns.
- Build a private knowledge base from business emails and archives.
- Create a retrieval tool that works entirely offline for sensitive projects.
Known limitations
Large file sets may require significant local disk space. Highly complex document layouts like nested tables or images may need pre-processing.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for LLM Ops.
Requires local storage for index files. Performance depends on local hardware specs. Best suited for text-heavy documents and code files.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 day ago
Permissions
No special permissions declared or detected