Knowledge base overview

Stars vs knowledge base boundaries, capability map, and onboarding path

What the knowledge base is

The knowledge base is a curated set of repositories you explicitly ingest for retrieval and Q&A in the RAG workspace. It is not the same as “all Stars.”

Stars vs knowledge base

	Stars	Knowledge base
Source	Synced from GitHub stars	Marked for ingest in Starcat
Scale	Can reach thousands	Curated (often dozens to ~200)
List search	✅	✅ (can also limit to knowledge)
Default RAG scope	❌	✅ ingested repos only
Typical index	FTS, etc.	Chunks + keywords + vectors

Independence (remember this):

• Starring does not auto-ingest
• Unstarring does not auto-remove from the knowledge base
• Not starred but ingested: still searchable; you can also subscribe to Releases

What you need

1. Pro
2. Chat + Embedding → AI configuration
3. Some ingested repositories → Adding repos

Capability map

I want to…	Go to
Add repos to the knowledge base	Adding repos
Start asking questions	RAG workspace (⇧⌘K)
Inspect cited source text	Citations & evidence
Manage session history	Session management
Fix low index coverage	Index management

Recommended onboarding path

1. Ingest 20–50 repos

   Prioritize repos you actively use or consult often.
2. Confirm Embedding and coverage

   Settings → AI + workspace Index tab.
3. Try @ one repo first

   Verify citations open correctly.
4. Then ask cross-repo questions

   Remove overly narrow @, or @ two or three repos to compare.

Entry points

• Sidebar Manage → Knowledge base / Library
• Toolbar “Knowledge base Q&A” or default ⇧⌘K
• Knowledge-related Smart Collections

Related

• Semantic search (list-level vector retrieval, not workspace Q&A)
• AI overview