How-to guides¶
Task-oriented recipes — you have a goal, these give the steps. They assume you already know what DocGen is; if not, start with the tutorial or the product brief.
| Guide | Task |
|---|---|
authoring-and-config-api.md |
Create, configure, export & update templates and documents over /v1 — the two paths (granular API + spec apply), endpoint reference, spec shape, worked flows. For humans and agents. Start here to drive DocGen. |
seeding-and-platform-tools.md |
Seed orgs/templates from specs, rebuild the databases (Postgres docgen via DATABASE_URL; SQLite demo.db fallback), and use the per-org platform tools. |
mcp-transport-and-library-enforcement.md |
Wire MCP source servers (FastMCP streamable-http), the fake DealDesk server, MCP_*_URL, and the library-enforcement CI. |
mcp-per-user-auth.md |
Forward end-user identity through DocGen to MCP source servers (per-user data scoping). |
observability-and-secrets-setup.md |
Set up Arize/Phoenix tracing and the config-vs-secret model when wiring a new project. |
onboard-a-solution.md |
Add a new org/client solution: fit check, solution docs, template specs, seeding, sources, and validation. |
Understanding how rendering and synthesis work (not a task) lives in
explanation/rendering-and-prompts.md. The exact interface — endpoints, spec schema — is underreference/(live schema:GET /v1/config-schema).