v1.12.4Friendlier Test button errors

AI Canvas Workflows,
InsideObsidian.

Turn Obsidian Canvas into an agent-ready AI workspace for generating images, videos, text, and audio from connected nodes.

Install in Obsidian
Workspace.canvas

A stunning medium-shot photograph of a young woman with vibrant red hair styled in a loose bun and bright red lipstick. She has fair skin and a slender, athletic physique. She is wearing a simple white bikini and a white bathrobe draped loosely over her shoulders,

Change the female to a muscular man, wearing sunglasses.

Generate a lyric for this picture.

The sun dips low, painting gold on the waves, a moment of silence is all that he craves.

Run
100:13
1
2

Men@img1 and women@img2 take turns posing against the railing for photos, holding the camera, with slight shaking. Use @audio as the bgm.

We trusted by creators from

Your Models, Your Providers.

Switch 25+ models across 13 providers for images, video, text, and audio — swap or fall back anytime without rebuilding your Obsidian workflow.

Image

Text-to-image, image-to-image, chain reference images upstream.

  • Nano Banana Pro
  • Nano Banana 2
  • GPT Image 2
  • Midjourney v8
  • Seedream 5.0
  • Grok Imagine
  • Seedream 4.5
  • Midjourney niji 7

Video

Text-to-video, first-frame, first/last-frame, image-ref, or extend.

  • Veo 3.1
  • Seedance 2.0
  • Kling 3.0
  • Veo 3.1 Lite
  • Seedance 2.0 Fast
  • Kling 2.6
  • Grok Video

Text

Visual reasoning, rewrites, planning. Split a response into many nodes.

I'm Claude, how can I assist you today?

  • GPT-5.4 Pro
  • Claude Opus 4.7
  • Gemini 3.1 Pro
  • Claude Sonnet 4.6
  • GPT-5.4
  • Gemini 3 Flash
  • Qwen 3.6 Plus

Audio

TTS, music, sound effects, dubbing, cover, isolation, transcription.

0:00 / 0:00
  • ElevenLabs TTS v3
  • ElevenLabs Music
  • MiniMax TTS
  • MiniMax Music
  • ElevenLabs Sound Effects

Providers

Bring your own API keys and connect multiple providers. Switch or fall back whenever one model is unavailable.

  • fal.ai
  • OpenAI
  • Anthropic
  • Google Gemini
  • Volcengine
  • ElevenLabs
  • AWS Bedrock
  • Kling
  • MiniMax
  • BytePlus
  • TokenRouter
  • Legnext
  • APIMart

Let agents drive your canvas.

Bragi supports exposing your Obsidian Canvas through a local MCP server. Agents can create nodes, connect ideas, run generations, and organize workflows for you.

MCPcodex mcp add bragi-canvas --url http://127.0.0.1:17775/mcp
Skillscodex skill install https://github.com/bragi/skills

Small Tools, Faster Workflow.

Thoughtful utilities reduce repetitive steps and keep your canvas organized, portable, and ready for the next idea.

Pan, mirror,
capture.

Open any equirectangular image as a true panorama. Drag to look around.

Split layouts.
Grid the canvas.

Tile nodes into a √N grid — compare variants or lay out storyboards.

Four in parallel.
Or drag-select all.

Run one prompt as four variants, or one model on every prompt at once.

Run (×4)

Your keys stay local.

Every API key lives in your vault's own config — never proxied, never synced to a server. You hit OpenAI, Anthropic, Gemini and friends directly.

Share a canvas as one file.

Pack canvas + every asset into a single .bragi bundle. Drop it into someone else's vault as a new canvas, or merge it into an existing one.

Clean house whenever.

One command sweeps every unreferenced media file out of the vault. Run it mid-experiment, keep the graph, reclaim the disk.

Your AI studio
inside your vault.

Create, automate, and share from the vault you already use.