with one click
start
// Zero-setup quick demo. Accept pasted content or a natural request and produce a useful structured artifact in under 5 minutes.
// Zero-setup quick demo. Accept pasted content or a natural request and produce a useful structured artifact in under 5 minutes.
[HINT] Download the complete skill directory including SKILL.md and all related files
| name | start |
| description | Zero-setup quick demo. Accept pasted content or a natural request and produce a useful structured artifact in under 5 minutes. |
| user-invocable | true |
| triggers | ["/start","try TARS","show me what TARS can do","quick demo","first session before /welcome has run"] |
| help | {"purpose":"Zero-setup demo flow. Detect what the user pasted, produce one useful structured artifact, save nothing unless the user opts in, then offer setup for compounding value.","use_cases":["Try TARS without setting it up","Show me what this can do with a meeting transcript","Quick test before onboarding"],"scope":"demo,onboarding,quick-start,evaluation"} |
Use this skill when the user wants to try TARS before committing to setup. The goal is visible value in one interaction, not a tour.
Check whether _system/config.md and _system/install.yaml exist.
_system/config.md is missing, or tars-user-name is empty, treat the user as new._system/install.yaml exists, read workspace_type, workspace_path, obsidian_enabled, and obsidian_vault_path so the closeout can describe persistence accurately./start. Otherwise route to the canonical skill that fits the content.If the user pasted content with the command, classify it immediately.
If no content was pasted, ask:
"Paste anything, a meeting transcript, email thread, design doc, customer call, or sales discovery notes. I'll show you what TARS does with it. Or describe what you want to try and I'll suggest a paste-target."
If the user has nothing handy, offer:
examples/pm-customer-call.mdexamples/eng-design-discussion.mdexamples/sales-discovery-call.md| Detected content | Preview behavior |
|---|---|
| Speaker labels, timestamps, or Zoom/Teams/Otter/Fireflies headers | Meeting summary, decisions, risks, action items |
| Email thread or Slack-style export | Follow-up draft plus memory candidates |
| RFC, design doc, ADR, or decision memo | Decision summary and durable memory candidates |
| Roadmap, launch, or status doc | Initiative status preview |
| Sales discovery or customer call notes | Call summary plus follow-up email draft |
| Generic notes or prose | Memory candidates and what TARS would remember |
Do not write to the workspace. Do not call mutation tools. Do not create notes. Do not update frontmatter. Default behavior is preview-only.
Produce the most useful artifact inline:
Cap output at about 2,000 words. If the content is large, summarize the visible preview and say that saving can preserve the full artifact after setup.
End with this block, filling in counts from the preview:
---
## What just happened
You ran TARS with zero setup. With `/welcome` (about 5 minutes), the same paste can also:
- Save {N_CANDIDATES} memory items so they show up in future `/briefing` and `/answer` results
- Extract {M_TASKS} action items into your TARS task notes, or into an external task system after you connect one
- Link people, decisions, and initiatives across future sessions
TARS starts as a local Markdown workspace in Claude. Obsidian is optional: turn it on later with `/welcome --enable-obsidian` if you want a visual note browser and `.base` views.
## What's next
1. **Set up TARS**: run `/welcome`
2. **See more**: run `/help`
3. **Save this demo**: say "save this" after setup to keep the artifact in your workspace
If the user replies "save this" or equivalent:
/learn or the canonical skill that created the preview and ask for confirmation before writing./welcome first, then repeat "save this".Never save by default.