VEYNDO
Handbook
docs/documentation-site.mdView source ↗

English source · 6 min

The connected Veyndo handbook

The public handbook follows a deliberate reading path: understand and try the product → connect intelligence → shape experiences → build apps and automation → operate commerce → control changes → inspect the foundation. It opens on the Veyndo website and uses the repository's Markdown as its source.

Find the right starting point

Your goal Start here
Understand what Veyndo does Feature guide and connected cognition
See what is online and actually tested Current status: one entry point for the deployed version, observed journeys and remaining work
Try a complete working round First tour
Give an agent a bounded job Agent workshop
Create an extension Visual App Studio and five illustrated workshops
Design a connected shop Experience integration, Storyfront and checkout
Deploy or assess the boundaries Architecture, hosting and security

Every rendered article provides its original source, chapter navigation and an article outline. Previous/next links follow the chapter sequence. Wide tables and code scroll within the article; the chapter menu collapses on small screens.

Search covers the complete Markdown inventory, including module READMEs and historical evidence. The curated chapters lead with current guides; the complete reference retains other sources and their original dates and scope.

One publication, two public surfaces

The existing Pages workflow builds the complete source inventory, search index, navigation, assets and canonical metadata. scripts/check_site.py checks every local link, anchor, article, search record and sitemap entry before publication. site/docs-navigation.json owns the curated chapter order; site/docs-copy.json owns the interface text.

The site package copies media used by the rendered pages and the screenshot viewer. Public manifests and downloadable projects keep their established URLs. The film guide publishes WebVTT for browser playback and copies linked SRT subtitles into the same site package for download. The site check compares those downloads with their source files and verifies the published links. Older source files are kept in the repository archive.

The private website runtime serves /docs/ on the apex from the fixed public GitHub Pages publication. It does not copy articles into another knowledge store or use merchant sessions. Cookies, authorization and tenant headers are never forwarded upstream. Only public documentation paths and GET/HEAD are accepted; recordings use streaming and bounded ranges. API, shop and Experience hosts keep their own existing routes. GitHub Pages remains an independently readable mirror.

On 11 November 2026, the compatibility page was updated to route the five old GitHub Pages prefixes (veyndo, vendune, rust-ai-commerce, sellm, and sellama) to the organization Pages mirror. Browser checks preserved the full article path, language query and section anchor. Requests to the old Vendune root, www, Team, start, Experience and Lumen shop hosts preserved their paths and queries in HTTPS redirects to the current hosts.

The old personal GitHub Pages site uses a browser redirect from its 404 page. For direct media requests and clients without JavaScript, use the current Veyndo handbook or its Pages mirror.

The site package now contains the four current SRT downloads alongside the WebVTT captions. After publication, 154 media responses matched the source bytes or, for videos and archives, a 1 KiB range and the original total size. The four current MP4 files also matched their full SHA-256 hashes on both hosts. The handbook browser played all four and opened the first SRT download from its visible link. Film guide.

Language and evidence

Handbook navigation, search messages and chapter descriptions ship in English, German, French and Spanish. The German platform overview is written in German. Technical guides are primarily in English. The handbook labels each article's source language in navigation and search. Set <!-- veyndo-language: de --> on a German article so that its metadata and browser language reflect the text.

The homepage and README use hosted browser captures for the Product Guide and checkout. The film index lists the dated tours and their scope. Older recordings remain available with their original evidence dates.

Build and validate

From the repository root, use a virtual environment if your Python installation requires one, then run:

python3 -m pip install -r site/requirements.txt
python3 scripts/build_site.py
python3 scripts/check_site.py
python3 scripts/testing/source_inventory.py
python3 scripts/api_catalogue.py --check
python3 scripts/structure.py

The structure check also needs npm ci in frontend/. The site renderer uses the pinned Python-Markdown dependency; ordinary commerce does not need it. Output is ignored .site/. Preview it with a local static web server.

The site check covers every source document, rendered page, local link, image, heading target, unique title, discovery metadata, search inventory and sitemap. Missing repository targets stop the build; missing rendered anchors fail the check. Untracked, non-ignored Markdown is included for local previews before its first commit; only committed files reach CI. Do not put private notes in the public repository.

Publication and current facts

Start with Status, the component map and the feature guide for the current product. The dated entries below describe earlier documentation releases.

The 21 October update consolidates current release and acceptance information in Status. Dated deployment notes move to the linked history; app and configurator guides link the specific public purchasing evidence. Read the current status first, then a dated receipt when its exact test scope matters. Build checks also cover all archived Markdown, screenshots and links.

The 16 October app review reconciles shared storage, saved agent work and enforced database/browser authority across README, app guides, architecture notes and the release tracker. App-system status is the reader's entry to delivered/partial/open behavior. The app chapter now links that map, the business-capability direction and the exact storage guide before deeper tutorials. Dated receipts remain unchanged; later results explicitly supersede earlier observations only for the later source tree.

The 12 October update adds five illustrated App Studio workshops and twelve fresh English browser captures. It reconciles the README/app references with package ownership, native controls, the 280-route catalogue and 424 Rust modules. Historical captures, measurements and public rollout observations retain their dates; the tutorials distinguish sample design, private preview, staging installation and actual clock/provider delivery.

The existing GitHub Pages workflow builds on every pull request and main push, and publishes only main or an explicit manual run. There is no path filter: new Markdown and code-derived module notes cannot silently miss a publication. GitHub's required verify check still protects merges. No authentication or production commerce setting changes are needed for documentation publication.

The October 6, 2026 review checked the repository's Markdown inventory, current module ownership, source links and rendered anchors. It corrected PostgreSQL/Qdrant versus legacy AGE/pgvector descriptions, enabled content languages, channel settings/lifecycle, worker roles, personal credentials, payment scope, source inventories and verification counts. Dated benchmarks, captures and test results retain their original measurement scope; a documentation review does not rerun every historical experiment or verify live provider accounts.

The 8 October review reconciled README, the full feature tour, app/Experience/channel guides, the homepage and llms.txt with 706102f. It corrected obsolete main-channel pause restrictions, EUR-only wording, the API/formal inventories and manually installed Storyfront claims. Three unretouched public demo captures complement the retained 6 October recordings. Historical release record and capture provenance distinguish deployed behavior, isolated tests and remaining native/provider activation work.

For future changes, read the actual owning code/configuration and update its guide in the same change. Regenerate module-inventory.md with python3 scripts/testing/source_inventory.py --write when source modules change. Review local module README lists too. Passing link/build checks does not prove every prose claim; observed behavior, implementation scope and proposed work must stay distinct. Use the testing guide and formal scope for exact evidence boundaries.

The subsequent 8 October core review is documented in core hardening, with a rendered request/event diagram, all eighteen findings, deployment settings, regressions and remaining proof/operational boundaries. Older latency and screenshot measurements retain their original scope; they are not relabelled as hardening results.