skills.rest
kwojtasinski-repo
kwojtasinski-repoCOMMUNITY
Explore all skills by this author →
GitHub

rag-reindex-decision

Determines whether RAG configuration changes require full re-indexing, incremental ingest, or query-time reload.

Updated Nov 19, 2020
Search Tags:#embeddings#rag#configuration#vector search#qdrant#ingest#reindexing
One-click install
npx skills add https://github.com/kwojtasinski-repo/ECommerceApp --skill rag-reindex-decision-kwojtasinski-repo
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: rag-reindex-decision
Source: https://github.com/kwojtasinski-repo/ECommerceApp/tree/main/.github/skills/rag-reindex-decision
Command: npx skills add https://github.com/kwojtasinski-repo/ECommerceApp --skill rag-reindex-decision-kwojtasinski-repo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Changing RAG configuration files without knowing the re-indexing impact leads to either wasted compute on unnecessary full rebuilds or stale indexes that silently return wrong results. This Skill provides a decision matrix that classifies each config change before you edit anything. ## Core Features & Use Cases - Change Classification Matrix: Maps edits to rag-config.yaml, metadata-rules.yaml, multilingual-glossary.yaml, queries.yaml, and chunker/embedder settings to query-time, incremental, or force-full outcomes. - Concrete Commands: Supplies the exact ingest and restart commands for each change type, covering both the Python and .NET RAG pipelines backed by Qdrant collections. - Pitfall Warnings: Documents common mistakes such as skipping --force-full after metadata-rules.yaml edits or forgetting to mirror glossary changes to the .NET copy. - Use Case: Before editing chunker.max_tokens in rag-config.yaml, consult the matrix to learn a full rebuild is required, then run python tools/rag/ingest.py --force-full and verify with compare_queries.py. ## Quick Start Ask the assistant to check whether your planned change to a RAG configuration file requires re-indexing before you edit it.

Frequently Asked Questions about rag-reindex-decision

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I know if a RAG config change requires re-indexing?▼

Look up the changed file and setting in the decision matrix. Query-time changes like ranking weights or glossary entries need only a server restart, while chunker, embedder, or metadata-rules changes require a full rebuild with --force-full.

When should I run ingest with --force-full?▼

Run python tools/rag/ingest.py --force-full after changing metadata-rules.yaml globs or kind values, chunker max_tokens or overlap, or the embedder model. These alter chunk boundaries or the embedding space, so existing chunks must be re-embedded.

Does editing the multilingual glossary require re-indexing?▼

No, multilingual-glossary.yaml changes are query-time only. Restart the rag-python-http or rag-dotnet-http server to reload the file, and mirror the edit to the .NET copy to avoid asymmetric multilingual behavior.

Why do doc_kind filters return stale results after editing metadata rules?▼

Existing chunks keep their old doc_kind until re-embedded. Editing metadata-rules.yaml without running --force-full updates classification logic only for new ingests, so a full rebuild is required to apply the change.

What happens when the embedder dimension changes?▼

An embedder.dim change requires dropping the Qdrant collection entirely and then running a force-full ingest, because the vector dimension of existing points no longer matches the new embedding space.

Related Skills

Explore More Software Engineering Skills→
LangConfigLangConfig
55

langchain-agents

Configure LangChain agents with tool bindings, memory, and RAG integration.

Official
Advanced
box755box755

llamaindex

Builds document-grounded Q&A and RAG pipelines with LlamaIndex ingestion, indexing, and querying.

Community
Advanced
eprouvezeeprouveze

recall

Search memory stores via SQLite FTS5 index and return top 5 cited results.

Community
Advanced
SteveGJonesSteveGJones
41

kb-rebuild-indexes

Rebuild knowledge base shelf-indexes by detecting modified library files with content hashes.

Community
Intermediate