Generate inside Obsidian with Smart Templates Pro
Smart Templates Pro adds a Generate path to the same context-first Template Context modal.
Core builds and copies the structured prompt.
Pro lets you generate inside Obsidian, review the output, then copy, insert, regenerate, or create a note.
Use this when the full loop should stay in your vault:
current work -> context -> template -> generate -> review -> insert only after review
Core vs Pro output paths
The first half of the workflow is identical to Core:
- Start from the current note or selection.
- Add more context when the assignment needs it.
- Select one or more trusted Markdown templates.
- Add optional instructions or a finish line.
Then you choose the output path:
| Path | Edition | Best when | Proof |
|---|---|---|---|
| Copy prompt | Core | You want to paste the structured request into ChatGPT, Claude, Gemini, Smart Chat, or another tool. | A ready-to-run prompt is copied. |
| Generate | Pro | You want to produce and review output without leaving Obsidian. | Generated output is reviewed before insertion or note creation. |
Core creates value without Pro generation.
Pro is for the moment when you want generation and review to stay inside the vault.
Review before insertion
After generation, Smart Templates opens a dedicated review surface.
Use it to:
- inspect the generated output
- copy the output
- insert it into the current note
- create a new note
- change the model and regenerate
The review step matters because generated output is not trusted note content yet.
A safe workflow is:
- Generate.
- Review against the current note, context, and template.
- Insert only the parts that match your intent.
- Improve the note or template if the output reveals a missing constraint.
Why generate inside Obsidian?
Generate is useful when you want fewer handoffs.
Instead of building a prompt, switching to another app, pasting, copying the result back, and deciding where it belongs, you can keep the request and review loop in the same Obsidian workflow.
Use it for:
- drafts that should return to the current note
- summaries that should become new notes
- model comparisons against the same context and template
- repeated outputs where review happens in Obsidian anyway
- workflows where copied prompts create too much switching friction
Model control
Pro gives you two levels of model control:
- a Smart Templates setting for the default Generate model
- a per-request override inside the modal
That means you can keep one stable default, then switch models only when the job needs it.
Use the override when:
- the request needs a stronger model
- the current model misses the format
- you want to compare outputs against the same context and template
- you are testing a local or cloud model configured in Smart Environment
Provider-backed workflows use the context and prompt you choose to send through that workflow.
Quick workflow
- Open Smart Templates from the current note or selection.
- Add the smallest useful context.
- Choose one template.
- Add a short instruction, such as "focus on risks and next actions."
- Click Generate.
- Review the output.
- Copy, insert, create a note, or regenerate with a different model.
You know it worked when:
One generated output has been reviewed before it is inserted into a trusted note.
Good defaults
- Start from current work, not a blank prompt.
- Add context before choosing the template.
- Use one template for the first pass.
- Keep optional instructions short.
- Review before insertion.
- Improve the note or template if the output missed the target.
Best use cases
Current-note draft
Use Generate when the current note already owns the assignment.
Example prompt instruction:
Use the attached context and template. Return a concise draft that I can review before inserting.
New note from selected context
Use Generate when the output should become its own note.
Good for:
- meeting summaries
- project briefs
- risk reviews
- literature summaries
- decision memos
Repeatable structured output
Use Generate when you often rebuild the same output shape.
The template gives the form.
The context gives the evidence.
The review step keeps the trusted note under your control.
Troubleshooting
The output used the wrong context
Reduce the context or make the current note's goal more explicit.
If the assignment depends on evidence from other notes, build context first with Smart Context Clipboard or Smart Context Builder.
The output format drifted
Use a more explicit template, reduce optional instructions, and regenerate.
Trusted templates reduce format drift, but they do not guarantee perfect output.
The model failed or returned weak output
Try:
- switching the model in the request
- making the outcome clearer
- adding missing context
- simplifying the template
- using Copy prompt and sending to a different provider
I do not need generation inside Obsidian
Use the Core path instead: copy the prompt and paste it into the AI tool you already use.