
TinyScale
io.github.sanv725v0.10.0Updated Oct 11, 2026
Add waitlists, feedback widgets, changelogs and deploy checks to a SaaS app; claim with no card.
Overview
Lets a coding agent add TinyScale SaaS building blocks such as waitlists, feedback widgets, changelogs and deploy checks to an app.
- What it does
- TinyScale is a local MCP server that runs in your project directory and lets a coding agent create a TinyScale project and wire its embeds into your app. Products include a launch waitlist, an anonymous feedback widget, an approved changelog, a browser deploy check, an onboarding checklist, CSV import, a captured demo and a report on how AI answers describe your product. The agent verifies the integration with a real result, such as a feedback submission ID or a passing browser check, then opens a claim page so you keep the project on the Free plan.
- When to use it
- Use it when you are building a SaaS or web app and want an agent to add launch waitlists, feedback collection, changelogs, onboarding checklists or deploy verification without setting up accounts or payment first. It suits quick previews that you later claim and continue on the Free plan.
- Requirements
- Runs as a local stdio process, typically via npx tinyscale mcp, and requires Node.js 20 or later. It runs in your project directory and stores credentials in a git-ignored .tinyscale/agent-state.json file. It talks to the hosted api.tinyscale.io service; no account, card or API key is needed for the preview.
Installation
In SourceWeft
- Open TinyScale in the dashboard and add it to a workspace.
- Enable the server for the chats that should use its tools.
Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.
Other MCP clients
Follow the launch instructions in the repository.
README
TinyScale
Small SaaS building blocks your coding agent can set up, verify, and hand to you.
Ask your agent to "add a launch waitlist" or "add a feedback widget and a changelog". It creates a TinyScale project with no account and no card, wires the embed into your app, proves it works with a real result, then opens a claim page so you keep it on the Free plan.
Use it from a coding agent (MCP)
Add the local MCP server to your agent. It runs in your project directory and keeps credentials in
.tinyscale/ (git-ignored, never returned to the model).
Claude Code
Or install the plugin, which adds the MCP server plus a skill that teaches Claude the setup flow:
Cursor, Windsurf, VS Code, Codex and other MCP clients
Then ask: "Use TinyScale to add a launch waitlist to this app and verify it works."
Use it from the command line
Requires Node.js 20 or later.
Each product has an exact agent guide: agent.md · waitlist · changelog · activate · import · demo · cite
How it works
- Preview — the agent creates a project that lasts 72 hours. No sign-up, no card. Limits are small and hard-stop; nothing is charged.
- Verify — the agent proves the integration with a real result: a feedback submission ID, a synthetic waitlist signup, a passing browser check with a receipt.
- Claim — you open the claim page, sign in, and the same project, IDs and embeds continue on Free. Public-facing actions (accepting real emails, publishing release notes) need your approval.
Security
- Credentials live only in
.tinyscale/agent-state.json(mode 0600, git-ignored). Tool results never include them or the claim link. - The CLI only talks to
https://api.tinyscale.io(plain HTTP only for localhost development). - Widgets only run on exact HTTPS origins you configure; browser checks require proof you own the origin.
Links
tinyscale.io · Products · Guides · Issues
MIT licensed client. The hosted service has its own terms at tinyscale.io.
Source: README.md at commit c5a322a
Tools
0Version history
1- v0.10.0LatestOct 11, 2026

