documentlisted
Install: claude install-skill DATA-AI-XYZ/Tandem
# Tandem: document (Technical Writer hat)
Operate as **Technical Writer hat**. The user wants a coherent, shareable documentation set synthesised from what Tandem already knows about the project — no separate doc-writing pass required.
## Default doc set
Author **exactly these five markdown files**, one per document, using these verbatim names as the output filenames:
| # | Document | Output file |
|---|----------|-------------|
| 1 | Overview | `documentation/overview.md` |
| 2 | Getting started | `documentation/getting-started.md` |
| 3 | Architecture | `documentation/architecture.md` |
| 4 | Decisions (digest) | `documentation/decisions.md` |
| 5 | Features (& usage) | `documentation/features.md` |
All output files are written into the **`documentation/` folder** at the project root, one `.md` per doc. Do not create subfolders inside `documentation/` — flat layout.
## Sources (read before authoring)
Read the following in order, resolving paths against the project root. Treat a missing file as "not present" (note the gap in the relevant section) rather than throwing.
1. **`PROJECT-CONTEXT.md`** — canonical project identity: name, purpose, tech stack, audience, deployment. This drives the Overview and Getting started sections.
2. **Epics** (`_00-Project-Management/30-Epics/EPIC-*.md`) — strategic scope. Skim titles + `## In scope` sections.
3. **Features** (`_00-Project-Management/31-Features/**/*.md`) — feature-level capabilities. Drives the Features (& usage) doc.