@vaicli/vai-workflow-doc-freshness
v1.0.0
Published
Audit a knowledge base for stale documents. Gathers collection metadata, model information, sample documents, and cost estimates, then generates a freshness report with recommendations.
Maintainers
Readme
vai-workflow-doc-freshness
Knowledge bases accumulate stale documents over time. Without a freshness audit, users retrieve outdated information and lose trust in the RAG system. Teams need a way to identify potentially stale content and estimate the cost of re-embedding.
Install
vai workflow install vai-workflow-doc-freshnessHow It Works
- Gather metadata — Use
collections,models, andexplainto get collection stats, model info, and educational context - Sample and test — Search to sample documents and estimate re-embedding costs
- Report — An LLM synthesizes all data into a freshness report with recommendations
Execution Plan
Layer 1 (parallel): collection_stats | model_info | embedding_context | sample_docs | reembed_cost
Layer 2: freshness_reportExample Usage
vai workflow run vai-workflow-doc-freshness \
--input collection="knowledge_base" \
--input staleness_days=60What This Teaches
- Five parallel steps in Layer 1 — the most parallel workflow in the catalog
modelstool retrieves current model capabilities and benchmarksexplaintool retrieves vai's educational content about the shared embedding spaceestimatetool provides concrete cost figures for re-embedding- This is the only workflow using
modelsandexplain, completing 14/14 tool coverage
License
MIT © 2026 Michael Lynn
