Sourcey fits your pipeline.
Sourcey is a CLI with static output, so it drops into the systems you already run. Build docs in CI, retrieve the plain-text output from a LangChain chain, or make a docs build one node in an n8n workflow.
GitHub Action
Build docs in CIRun sourcey build on every push and deploy the static output to GitHub Pages, Vercel, or anywhere. The docs stay in lockstep with the source they came from.
View on GitHub →- uses: sourcey/build-docs@v1
with:
output: docs LangChain
Retrieve the docs an agent readsSourceyRetriever pulls llms-full.txt and the search index into a chain, so an agent grounds on the same documentation graph your site renders. Python and JavaScript.
View on PyPI →from langchain_sourcey import SourceyRetriever
r = SourceyRetriever(
site_url="https://yourdocs.com/docs",
) n8n
Docs in your automationsThe Sourcey community node loads and retrieves your docs inside n8n workflows, so a docs step can feed a vector store, a support router, or anything downstream.
View on npm →npm install n8n-nodes-sourcey
# adds the Sourcey node:
# Load all pages
# Retrieve by query Want Sourcey somewhere else? The CLI is a single command with static output, so it scripts cleanly into any pipeline. Open an issue at sourcey/sourcey.