Tighten Sourcey integration docs voice

Author: auscaster

README.md

@@ -4,9 +4,10 @@
 [![PyPI - License](https://img.shields.io/pypi/l/langchain-sourcey)](https://pypi.org/project/langchain-sourcey/)
 [![CI](https://github.com/sourcey/langchain-sourcey/actions/workflows/ci.yml/badge.svg)](https://github.com/sourcey/langchain-sourcey/actions/workflows/ci.yml)
 
-Build your own Ask AI on top of a published Sourcey docs site.
+Implement Ask AI over a published Sourcey docs site.
 
-`langchain-sourcey` is the retrieval layer behind that feature.
+`langchain-sourcey` reads Sourcey's generated search and LLM artefacts and
+returns canonical page URLs for citation.
 
 Sourcey already emits the files a retriever needs:
 
@@ -49,7 +50,7 @@ for doc in docs:
 
 For a runnable script, see [examples/live_quickstart.py](examples/live_quickstart.py).
 
-More context: `https://sourcey.com/docs/guides/guide-langchain-retriever`
+More context: [Sourcey guide](https://sourcey.com/docs/guides/guide-langchain-retriever)
 
 ## Implement Ask AI
 
@@ -90,9 +91,9 @@ print(answer)
 
 For a fuller example, see [examples/rag_chain.py](examples/rag_chain.py).
 
-## What Has To Exist
+## Sourcey Output Contract
 
-For clean retrieval, the published Sourcey site should expose:
+For predictable retrieval, the published Sourcey site should expose:
 
 - publish `search-index.json`
 - publish `llms-full.txt`

docs/langchain-js/provider-sourcey.mdx

@@ -5,8 +5,9 @@ description: "Build Ask AI over Sourcey docs using LangChain JavaScript."
 
 [Sourcey](https://sourcey.com) builds static docs sites.
 
-`langchain-sourcey` is the retrieval layer behind an Ask AI feature over a
-published Sourcey docs site.
+Use `langchain-sourcey` to retrieve from a published Sourcey docs site. It
+reads `search-index.json`, uses `llms-full.txt` when present, and returns
+canonical page URLs for citation.
 
 ## Installation and setup
 

docs/langchain-js/retriever-sourcey.mdx

@@ -5,9 +5,9 @@ description: "Integrate with the SourceyRetriever retriever using LangChain Java
 
 [Sourcey](https://sourcey.com) already emits the files this retriever needs.
 
-`SourceyRetriever` is the retrieval layer behind an Ask AI flow over a Sourcey
-docs site. It reads `search-index.json`, hydrates from `llms-full.txt`, and
-returns `Document` objects with canonical page URLs in `metadata.source`.
+Use `SourceyRetriever` to retrieve from a published Sourcey docs site. It
+reads `search-index.json`, uses `llms-full.txt` when present, and returns
+`Document` objects with canonical page URLs in `metadata.source`.
 
 ## Integration details
 

docs/langchain/provider-sourcey.mdx

@@ -5,8 +5,9 @@ description: "Build Ask AI over Sourcey docs using LangChain Python."
 
 [Sourcey](https://sourcey.com) builds static docs sites.
 
-`langchain-sourcey` is the retrieval layer behind an Ask AI feature over a
-published Sourcey docs site.
+Use `langchain-sourcey` to retrieve from a published Sourcey docs site. It
+reads `search-index.json`, uses `llms-full.txt` when present, and returns
+canonical page URLs for citation.
 
 ## Installation and setup
 

docs/langchain/retriever-sourcey.mdx

@@ -5,9 +5,9 @@ description: "Integrate with the Sourcey retriever using LangChain Python."
 
 [Sourcey](https://sourcey.com) already emits the files this retriever needs.
 
-`SourceyRetriever` is the retrieval layer behind an Ask AI flow over a Sourcey
-docs site. It reads `search-index.json`, hydrates from `llms-full.txt`, and
-returns `Document` objects with canonical page URLs in `metadata["source"]`.
+Use `SourceyRetriever` to retrieve from a published Sourcey docs site. It
+reads `search-index.json`, uses `llms-full.txt` when present, and returns
+`Document` objects with canonical page URLs in `metadata["source"]`.
 
 ## Setup
 

js/README.md

@@ -1,8 +1,9 @@
 # langchain-sourcey
 
-Build your own Ask AI on top of a published Sourcey docs site.
+Implement Ask AI over a published Sourcey docs site.
 
-`langchain-sourcey` is the retrieval layer behind that feature.
+`langchain-sourcey` reads Sourcey's generated search and LLM artefacts and
+returns canonical page URLs for citation.
 
 Sourcey already emits the files a retriever needs:
 
@@ -54,7 +55,7 @@ for (const doc of docs) {
 
 For a runnable script, see [examples/live-quickstart.ts](examples/live-quickstart.ts).
 
-More context: `https://sourcey.com/docs/guides/guide-langchain-retriever`
+More context: [Sourcey guide](https://sourcey.com/docs/guides/guide-langchain-retriever)
 
 ## Implement Ask AI
 
@@ -119,9 +120,9 @@ console.log(answer);
 
 For a fuller example, see [examples/rag-chain.ts](examples/rag-chain.ts).
 
-## What Has To Exist
+## Sourcey Output Contract
 
-For clean retrieval, the published Sourcey site should expose:
+For predictable retrieval, the published Sourcey site should expose:
 
 - publish `search-index.json`
 - publish `llms-full.txt`