Index and validate RAG corpus knowledge nodes using standard-library Python scripts.
npx skills add https://github.com/arbgjr/smart_alarm --skill rag-curator
Please help me install this Agent Skill. Skill: rag-curator Source: https://github.com/arbgjr/smart_alarm/tree/main/.claude/skills/rag-curator Command: npx skills add https://github.com/arbgjr/smart_alarm --skill rag-curator
The Rag Curator streamlines the management and quality control of the project knowledge base by automating the ingestion, indexing, and upkeep of RAG corpus nodes, ensuring consistent access to relevant information.
Use the rag-curator to index ADRs and maintain the corpus.
High-intent search queries and answers about installing and using this skill.
To index ADRs into a RAG corpus, run the Python script index_adrs.py using the --all flag. This copies architectural decision records into corpus/nodes/decisions for consistent retrieval across projects.
The best way to validate RAG corpus quality is using the curator's validation scripts. They check completeness and consistency of corpus nodes to prevent information gaps during retrieval.
Yes, you can use Python standard libraries to maintain a RAG knowledge base. The curator relies entirely on standard libraries to run scripts that index and validate corpus data without external dependencies.
To clean obsolete learnings from a RAG corpus, use the curator's cleanup scripts. They remove outdated or irrelevant knowledge nodes from corpus/nodes/learnings to keep the corpus current.
Yes, the curator supports indexing learnings for a specific project. You can target individual projects by passing the project ID argument to the Python indexing scripts.
Your RAG corpus returns outdated information because obsolete nodes are not cleaned. Run the curator's validation to check node completeness and remove outdated knowledge to keep the corpus current.
carson-sweetMove approved project documents to canonical storage with provenance tracking.
kevinnguyen271090Build and deploy AI agents with Google ADK Python.
adeboldPerform vector-based semantic search against an AgentDB index.
factorial-ioCurate session logs into structured knowledge nodes with interactive contradiction resolution.
