m3-doc-ai

name	m3-doc-ai
description	Translate and generate documentation for the monorepo with the @aphrody-code/doc-ai CLI (Google Gemini backend, offline fallback). Use when asked to "translate the docs", "generate an API guide for a Lit component", "sync docs to another language", or "run doc-ai".

m3-doc-ai

@aphrody-code/doc-ai is the doc-as-code CLI of the monorepo: it translates Markdown, generates Lit component API guides, and bulk-syncs every docs/ folder — backed by Google Gemini (via the Vercel AI SDK @ai-sdk/google + ai), with a deterministic static fallback when no credentials are set (so it stays runnable offline and in CI).

Commands

doc-ai translate <file_or_dir> [lang]   # translate one .md or a dir (recursive); default lang fr
doc-ai generate <ts_file> [out_file]    # generate a Lit component API guide from source
doc-ai sync [lang]                      # translate every monorepo docs/ folder
doc-ai --help

From the monorepo root the wrappers are wired in:

bun run docs:translate <file_or_dir> [lang]
bun run docs:ai                          # sync

Translations are written next to each source as <name>.<lang>.md (default fr) — this is why the repo carries *.fr.md siblings.

Configuration

Env var	Effect
GEMINI_API_KEY | GOOGLE_API_KEY	Enable Gemini-backed generation/translation.
DOC_AI_MODEL	Override the model (default gemini-2.5-flash).
DOC_AI_GEMINI_BASE_URL	Override the Generative Language API base URL.
DOC_AI_REQUEST_TIMEOUT_MS	Per-request network timeout (default 60000).

Without credentials, generate/translate fall back to a deterministic static template — the pipeline never hard-fails for lack of a key.

When to use

• A new docs/*.md was written in English -> doc-ai translate docs/new.md fr.
• A new md-* component shipped -> doc-ai generate packages/material-web/<comp>/internal/<comp>.ts to draft its API guide.
• Before a docs release -> bun run docs:ai to refresh every translation.

Never paste an API key into a tracked file; pass it via the environment only.

Develop in-repo: cd packages/doc-ai && bun run build && bun run test.