Personal Knowledge Base (RAG)

Pain Point

You read articles, tweets, and watch videos all day but can never find that one thing you saw last week. Bookmarks pile up and become useless. Your knowledge is scattered across browser tabs, chat histories, and note apps with no unified search.

What You Can Do

• Drop any URL into Telegram or Slack and it auto-ingests the content (articles, tweets, YouTube transcripts, PDFs).
• Semantic search over everything you've saved: "What did I save about agent memory?" returns ranked results with sources.
• Cross-workflow integration: Other workflows (e.g., video idea pipeline, meeting prep) can automatically query the KB for relevant saved content.
• Citation-backed answers: Every response includes source documents and relevant excerpts.

Skills You Need

• knowledge-base skill (or build custom RAG with embeddings)
• web_fetch (built-in) for fetching URL content
• Telegram topic or Slack channel for ingestion
• Vector store for semantic search

How to Set It Up

1. Choose Your Ingestion Channel

Create a Telegram topic called "knowledge-base" (or use a Slack channel). This is where you'll drop URLs for automatic ingestion.

2. Configure the Agent

Use the "Basic" prompt above. The agent will:

• Watch for URLs dropped in your channel
• Automatically fetch and chunk content
• Store embeddings with metadata (title, URL, date, type)
• Reply with confirmation of what was ingested

3. Query Your Knowledge Base

Ask questions in the same channel. The agent performs semantic search and returns top results with sources and relevant excerpts.

4. Connect to Other Workflows

The real power comes from cross-workflow integration. When other agents need research (video ideas, meeting prep, content creation), they can automatically query your knowledge base for relevant saved content.

5. Test It

Drop a few URLs and ask questions like "What do I have about LLM memory?" to verify everything works.

Tips

• Start with a single ingestion channel and expand later.
• The more you feed, the more valuable it gets — knowledge extraction compounds over time.
• Use metadata filtering to search within specific content types (articles vs. videos vs. PDFs).

Deploy with ShipClaw

Skip the setup — get a fully managed OpenClaw instance ready to run this use case.

Quick Start

1. Pick a plan — Pro recommended for this use case
2. Go to your Instances Dashboard and click Deploy New Instance
3. Once deployed, use the sample prompt above to configure your agent
4. Customize thresholds, schedules, and sources to fit your workflow