---
title: Jamdesk Workflow
description: How to run docs-as-code for Assistiv in Jamdesk, including what can be automated and what still needs authoring.
---

> **For AI agents:** the complete documentation index is at [llms.txt](/llms.txt). Append `.md` to any page URL for its markdown version.

This page answers two practical questions:

1. What does Jamdesk automate today?
2. What should our team automate next for full-stack Assistiv docs?

## What Jamdesk Automates Now

- Build and deploy on push to the connected branch
- Validation of `docs.json` and MDX pages during build
- AI-ready outputs like `llms.txt` and markdown-source page access (`.md` URL suffix)
- OpenAPI endpoint page generation when an OpenAPI spec is configured

## What Is Not Fully Automatic Yet

- End-to-end conceptual documentation generated from all repos with no guidance
- Full architecture narratives and business-context explanations
- Guaranteed accurate cross-repo change logs without a workflow layered on top

## Practical Automation Pattern for Assistiv

1. Keep this repo as the published source of truth.
2. Add a lightweight "docs update required" checklist to each product PR.
3. Use AI agents (Codex/Cursor) to draft MDX updates from code diffs.
4. Review docs changes in the same PR as code changes whenever possible.
5. Introduce OpenAPI specs for backend and Aurora to auto-generate endpoint docs.

## Codex-Based Automation Option

Jamdesk documents a Codex workflow using:

- project-level writing rules in `AGENTS.md`
- optional MCP endpoint config for docs retrieval
- optional `/update-jamdesk` skill for automated doc updates

This is currently the fastest path to partial automation without building a custom doc generation pipeline.

## Suggested Near-Term Improvements

1. Add an OpenAPI export step in CI for backend and Aurora.
2. Generate route inventory artifacts on each merge and include doc diff checks.
3. Add ownership metadata to docs pages so each section has a maintainer.