Repository navigation
Releases
Releases · wiroai/Wiro-Obsidian
1.1.1
A maintenance release for Obsidian's plugin review.
- On desktop, the writing commands still show the answer as the model writes it. Node's
httpsmodule now loads through a dynamicimport()inside aPlatform.isDesktopcheck, instead ofrequire(). On mobile nothing changes: the answer appears when it is complete. - The README's Network use and privacy section now says why the file fields of Run a model and Run a workflow list the vault's images, audio files and videos: so you can pick one. Only the file you pick is read and uploaded.
1.1.0
Wiro AI for Obsidian now covers images, video, audio, writing and any model in the Wiro AI catalog.
New commands
- Edit image and Upscale image work on an image that is already in the note and save the result under it.
- Generate video turns a prompt into an MP4 file embedded in the note.
- Read note aloud reads the whole note into an audio file at the top. Generate speech now has the voice and language in its dialog.
- Transcribe audio puts the transcript of an audio file, including a
.webmvoice recording, under it as a callout, a quote or a paragraph. - Generate music makes a track from a description.
- Summarize note or selection, Rewrite selection, Translate selection, Continue writing and Ask about this note use the Wiro AI LLM gateway. On desktop the text appears as the model writes it. Pick the model under Settings → Wiro AI → Text.
- Run a model opens the catalog by kind, with search and prices, and builds a form from the inputs of the model you pick.
- Run a workflow runs the workflows of your Wiro AI project through a form and shows the step it is on.
- Open history lists the runs from this vault with their model, cost, files and note, plus your latest tasks on Wiro AI.
Changes
- Every run dialog shows the price for your options before it runs, and the button repeats it.
- The settings have one page per kind of model, with a list to start from, Browse the catalog, the price and the model's options.
- Settings from 1.0.0 carry over: the image and speech models, their options, extra parameters and folders.
Network use
The writing commands talk to llm.wiro.ai, and Edit image, Upscale image, Transcribe audio and the file fields of forms upload the file you choose to Wiro AI's file storage. The README's Network use and privacy section lists every request.
1.0.0
First release of the Wiro AI plugin for Obsidian.
- Generate image: type a prompt, or start from the selection, and the image is embedded after the current line.
- Generate image from selection: the selected text becomes the prompt and the image goes under it.
- Generate speech and Generate speech from selection: text to an audio file, embedded in the note. Markdown is read as plain text.
- Right-click a selection for both actions.
- Nine image models (Nano Banana 2.1 by default) and five speech models (Eleven v4 by default) are built in, and any model ID from wiro.ai/models works too. The settings show each model's own options, such as aspect ratio, resolution and voice.
- A progress notice with a Cancel button while the task runs, and the cost of the run when it ends.
- The API key stays in Obsidian's keychain, outside the vault.
Needs Obsidian 1.13.2 or later and a Wiro AI API key from wiro.ai/panel/project/new.
main.js, manifest.json and styles.css are built by the release workflow and carry a build provenance attestation.