Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →jeremylongshore/langchain-embeddings-search
Build and query vector stores with LangChain 1.0 without getting burned by flipped score semantics, embedding-dim mismatches, reranker quirks, and chunk-splitter bugs.
Passed 20 of 21 test cases with the skill, and 15 of 21 without it.
and 2 more in the full report
5 fixed16 passed either way1 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 4 test cases the skill fixed. Same model, same prompt, with and without the skill.
I am building a Python RAG pipeline using LangChain to index technical Markdown documentation containing code snippets. Could you write a complete setup script that loads and chunks the Markdown safely without breaking code blocks, initializes OpenAI embeddings with a startup dimension assertion, sets up vector retrieval with normalized score outputs, and combines vector search with keyword matching via an ensemble retriever?
| Checked | Without skill | With skill |
|---|---|---|
| Uses RecursiveCharacterTextSplitter.from_language with Markdown language support | Passed | Passed |
| Asserts that embedding vector length equals 1536 at startup | Passed | Passed |
| Transforms FAISS L2 distance scores using 1.0 divided by (1.0 + score) | Missed | Passed |
| Combines BM25 search and vector search using EnsembleRetriever | Passed | Passed |
The instructions your model reads when the skill loads.
FAISS.similarity_search_with_score() returns L2 distance — lower is better.
Pinecone.similarity_search_with_score() returns cosine similarity — higher is
better. Swap your vector store and your if score > 0.8 filter now keeps the
garbage and drops the good results, silently. This is pain-catalog entry P12,
and it is the single most common reason a "we migrated from FAISS to Pinecone
for scale" project loses retrieval quality overnight.
The sibling gotchas:
RecursiveCharacterTextSplitter default separators break inside code
fences, so RAG over Markdown docs truncates code examples mid-functionVectorStore.__init__; the failure blames "dim
mismatch: 1536 != 3072" and no earlier errorThis skill walks through embedding model selection, vector store creation with
the version-safe dim guard, score normalization, hybrid keyword+vector search,
and rerankers with the correct filter-by-rank pattern. Pin: langchain-core 1.0.x,
langchain-community 1.0.x, langchain-openai 1.0.x, faiss-cpu, pinecone-client.
Pain-catalog anchors: P12, P13, P14, P15, P49, P50.
langchain-core >= 1.0, < 2.0 and langchain-community >= 1.0, < 2.0pip install langchain-openai (text-embedding-3-small/large)pip install faiss-cpu OR pip install langchain-pineconeOPENAI_API_KEY, PINECONE_API_KEYtest cases passed, vs 15 / 21 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.