stnd.buildSTANDARD MANUAL2026-07-15

5.3 Core API Reference

Reference specifications for the Standard Garden HTTP API.

5.3 Core API Reference

The Standard Garden API lets external clients — the Obsidian plugin, the CLI, your own scripts — publish and manage notes in your garden.

Authentication

All endpoints require an API key, sent as a header:

x-api-key: sg_...

Get your key via the ::api command in the Command Palette (sign in first). The same view lets you rotate the key and copy it. Keys are stored hashed (SHA-256) — the plaintext is only shown at generation, so copy it then.

Endpoints

GET /api/me

Verify a key and identify its owner. Used by clients to confirm the connection.

Response:

{ "username": "jean", "id": "..." }

PUT /api/publish/:slug

Create or update a note. The slug is the identifier — publishing twice to the same slug updates the note (idempotent, perfect for one-way sync from Obsidian).

Body:

{ "content": "# My Note\n\nMarkdown content…", "title": "My Note" }

Response: 201 (created) or 200 (updated)

{ "success": true, "url": "/@jean/my-note", "slug": "my-note", "nano_id": "..." }

Rate limit: 60 requests/minute — roomy enough to sync a large vault.

DELETE /api/publish/:slug

Unpublish a note. Removes it permanently, including its attachments.

Response: { "success": true, "slug": "my-note" }

POST /api/publish/attachment

Upload an image before publishing the note that embeds it.

Format: multipart/form-data with fields file (the image) and slug (the note it belongs to).
Limits: images only (png, jpeg, gif, webp, svg, avif) · 200 MB max per file · 30 requests/minute.

Response:

{ "url": "/cdn/…", "filename": "uuid.png" }

POST /api/ai/theme

Generate design tokens for a note — the AI reads the content (and an optional style instruction) and returns frontmatter tokens (colors, fonts, rhythm). This is what powers the “Generate” button in the Obsidian plugin’s panel.

Body:

{ "instruction": "warm autumn palette", "noteContent": "…", "currentTokens": {} }

Response: { "tokens": { "color-light-accent": "#…", "font-header": "…" } }

Rate limit: 10 requests/minute (each call costs real AI tokens).

Errors

All endpoints return JSON errors with meaningful status codes: 401 (missing/invalid key), 400 (bad input), 413 (file too large), 429 (rate limited — check the Retry-After header), 503 (service not configured).

See Also