Search indexed knowledge chunks with optional source filters and ranked attribution.
npx skills add https://github.com/mufarosean/Parallx --skill search-knowledge
Please help me install this Agent Skill. Skill: search_knowledge Source: https://github.com/mufarosean/Parallx/tree/main/.parallx/skills/search_knowledge Command: npx skills add https://github.com/mufarosean/Parallx --skill search-knowledge
Semantic search across all indexed knowledge (canvas pages and workspace files) to surface the most relevant chunks with source attribution when context is insufficient.
Search your indexed knowledge by entering a natural language query to retrieve the most relevant chunks with attribution.
High-intent search queries and answers about installing and using this skill.
Semantic search across canvas pages and workspace files is performed by entering a natural language query to retrieve the most relevant knowledge chunks ranked via a vector-embedding pipeline.
Vector-embedding search for indexed knowledge finds relevant text chunks by matching query semantics, returning ranked results with source attribution to locate information across canvas and workspace contexts.
Yes, you can filter semantic search results by applying a source_filter to constrain the search to either page_block or file_chunk contexts within the indexed knowledge.
Source attribution for indexed knowledge chunks works by tagging each ranked search result with its origin context, allowing you to trace retrieved information back to the specific canvas page or workspace file.
The best way to find specific information across canvas and workspace contexts is running a natural language semantic search to surface relevant chunks with source attribution when context is insufficient.
Yes, canvas pages and workspace files must be indexed before running a semantic search, because the vector-embedding pipeline relies on indexed knowledge chunks to match and rank natural language queries.
mindmorassIngest PDFs, websites, code repos, and notes into vector databases with chunking and metadata schemas.
daffy0208Implement Retrieval-Augmented Generation pipelines with vector stores and hybrid search.
1thirteeng3Create serverless Pinecone indexes and query embedding vectors.
zilliztechCombine BM25 keyword matching with vector semantic similarity in Milvus.
