knowledge-base-search
Search and retrieve information from the vector knowledge base (RAG)
Installation
Suivez les instructions du dépôt pour votre agent. Consultez les fichiers et dépendances avant l’installation.
Voir la source ↗GuidesInstructions
Knowledge Base Skill
You have access to a vector knowledge base that stores documents and enables semantic search. This knowledge base uses embeddings to find contextually relevant information.
Capabilities
-
Semantic Search: Search for documents using natural language queries. The search uses vector similarity to find contextually relevant content, not just keyword matching.
-
Collection Management: Documents can be organized into collections. Use collection filters to narrow your search to specific domains or topics.
-
Context Retrieval: Retrieve relevant context from documents to help answer user questions. Multiple document chunks may be combined to provide comprehensive information.
When to Use This Skill
Use the knowledge base when:
- The user asks about specific topics that may be in uploaded documents
- You need factual information from the user's document collection
- Looking for specific details, quotes, or data from stored documents
- The user references documents they've previously uploaded
Search Guidelines
-
Be Specific: Use clear, focused queries. Instead of "tell me about the project", try "project requirements and timeline".
-
Iterate: If initial results aren't helpful, rephrase your query or try different keywords.
-
Use Context: Combine multiple search results to build a comprehensive answer.
-
Cite Sources: When using information from search results, mention the source document.
Example Workflows
Finding Specific Information
- Call
kb_searchwith a targeted query - Review the returned chunks for relevant information
- If needed, search again with refined query
- Synthesize information from multiple results
Exploring Available Documents
- Call
kb_list_collectionsto see what's available - Call
kb_list_documentsto see specific files - Search within relevant collections for specific content
Checking System Status
- Call
kb_get_statsto verify the knowledge base is available - Check document and chunk counts to understand the knowledge base size
Exemples
Consultez les exemples du dépôt source.
Voir la source ↗Historique source
- Date d’indexation
- 2026-01-20
- Mise à jour de l’index
- 2026-01-26
- Branch
- main
Les dates concernent ce répertoire. Consultez le dépôt pour l’historique du code.
GitHub ↗chat Comments (0)
Sign in to join the discussion and leave a comment.
No comments yet. Be the first to share your thoughts!