App scaffoldsbrief → artifacts
Spec a feature, get an agent-ready brief
“Plan a multiplayer lobby on Firebase: schemas, latency budget, phased rollout.”
4 markdown filesLine chartPhase table
Studio · Workspaces & run artifacts
Ship specs your coding agent can execute.
Xenonflare Studio queues work from a short brief. When generation completes, you review files, charts, and tables in the dashboard — everything organized per workspace, ready for Cursor, Copilot, or any agentic flow.
- Passwordless magic-link sign-in
- Per-workspace chat threads
- Charts + tables alongside files
- Self-host with one Docker image
studio.xenonflare.comLive
Workspace · ProductSpec
Draft a real-time multiplayer lobby on Firebase. Include schemas, latency budget chart, and a phased rollout table.
Queued. Running on runner-7. Producing 4 files, 2 charts, 1 table…
generating…
Artifacts
Frontend.md●
Backend.md○
Schemas.md○
- 6 chart types
- bar · pie · line · area · scatter · stacked
- Multi-file
- Markdown bundles per run
- Parallel
- Self-host extra runners
- Stripe
- Plans + customer portal
What you can build
One brief in. A workspace of artifacts out.
The studio is good at structured outputs: things you would normally split into a doc, a chart, and a spreadsheet — produced together.
Researchbrief → artifacts
Distill noisy inputs into a clean summary
“Compare 6 vector DBs for a 50M-doc workload. Score on cost, latency, ops burden.”
Comparison tableScoring chartRecommendation file
Roadmapsbrief → artifacts
Turn ideas into a phased plan
“Quarterly roadmap for a B2B analytics app. Show effort vs. impact and a checklist per phase.”
Scatter chartEffort tableChecklist artifacts
Internal docsbrief → artifacts
Bootstrap onboarding & runbooks
“Write an on-call runbook for our auth service: incidents, dashboards, escalation tree.”
Multi-file docsPie chartEscalation tree (SVG)
Product surface
One results view: prompts, visuals, grids, and account controls.
Charts, files, tables, billing — plus an open-source runner you can self-host. Details live in the docs.
Charts that explain the run
Bar, pie, line, area, scatter, and stacked-bar charts ride alongside prose — quick sanity checks before you ship specs to an agent.
Dataset tables
Structured tables (datasets) for scores, comparisons, and checklists — easy to scan, easy to copy.
| Phase | Tokens | Status |
| Plan | 1.2k | OK |
| Build | 8.4k | OK |
| Review | 2.1k | … |
| Ship | 4.6k | OK |
Multi-file outputs
Each completion is a small library of markdown — Frontend, Backend, agents, and more. Copy one file or the whole set.
Frontend.md●
Backend.md○
Agents.md○
Roadmap.md○
Billing & plans
Upgrade with Stripe, open the customer portal for invoices, and keep terms acceptance in sync with checkout.
Pro · monthlyStripe
Daily tokens · 62%Resets 00:00 UTC
Flow
Three steps. No credits spreadsheet.
- 01
Describe the build
Paste a product idea, stack hints, and constraints — we queue a structured job tied to a workspace thread.
- 02
Generate on your hardware
Runners pick up the job in order. Use the shared pool on Free or self-host with your own API key for full throughput.
- 03
Review in the studio
Open results: skim charts and tables, copy per-file prompts, and paste into Cursor or any agentic workflow.
Why this shape
Cloud queues. Your hardware generates.
Tiers
Free, Plus, and Pro
Daily token credits per tier; compare on the pricing page, then subscribe in Settings → Billing.
Privacy
Keys stay local
Model calls run on hardware you control. The cloud only queues work and stores outputs for review.
Queue
Fair, transparent ordering
Work is processed in order with clear states from queued through complete — no mystery inboxes.
Scale
Bring your own throughput
Run more capacity on infrastructure you control so queued work finishes faster — no shared credentials across hosts.
Self-host
Run the open-source worker on your own box.
The cloud queues work and stores results. The model call happens in a small Node worker you control — your API key never leaves the host. Spin up more processes to drain the queue faster.
API keys stay on your hardware
Same lease/heartbeat endpoints
Scale by adding processes
One Docker image, env-driven
~/runner
$ git clone Xenon-Flare/runner
$ export RUNNER_TOKEN=…
$ export OPENAI_API_KEY=…
$ npm start
[runner-7] connected
[runner-7] leased ws_4Q9a · 12.4k tok
[runner-7] complete · 4 files · 2 chartsYour next repo starts as one good brief.
Passwordless login, dedicated workspaces, and a results surface built for builders — not slide decks. Queue a run, skim charts and tables, then iterate until the agent output feels right.