muse-marketplace
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Muse Marketplace
The open marketplace & registry for shareable Muse assets.
Muse users will soon be able to share Muse Avatars with each other — an avatar sharing pop-up, animated previews of avatar "modes" like Working, the works. But sharing needs somewhere to live: an open registry where anyone can publish an avatar, persona, voice, or bundle, where every file carries a checksum, and where "safe" means actually checked, not just claimed.
That is this repo. We are building the Muse Avatar marketplace — starting now, in the open.
npx @reneturcios/muse-marketplace list
npx @reneturcios/muse-marketplace search tree
npx @reneturcios/muse-marketplace show jolly
npx @reneturcios/muse-marketplace install jolly
npx @reneturcios/muse-marketplace doctorOr browse it as a site: open site/index.html (or the GitHub Pages deploy) — a dark, searchable grid with an animated mode switcher on every avatar.
What's inside
registry/— The registry itself. Every listing is a folderregistry/<kind>s/<slug>/with alisting.json+ its asset files. Kinds:avatar(primary),persona,voice,bundle.registry/index.jsonis generated from the listings (muse-marketplace index), never hand-edited; CI fails on drift.Launch catalog (9 listings, all original, all installable) — 6 avatars: Jolly (the friendly green companion), Tree (a forest spirit), Straw Hat Explorer, Neon Founder, Cozy Scholar, Pixel Sprite — each with an original animated SVG per mode (idle / working / thinking / celebrating, plus extras like making, dance, milestone, error). 2 personas: Sharp Operator, Warm Companion. 1 bundle: Starter Pack (everything).
CLI + MCP in one bin —
muse-marketplace(aliasmmkt) for humans; the same binary is an MCP server (muse-marketplace mcp) exposingmarketplace_search,marketplace_list,marketplace_show,marketplace_validate,marketplace_install.site/— A zero-build static marketplace: search, filter by kind + tag, click a card for the animated mode switcher and a copyable install command.docs/— SPEC (the listing schema), PUBLISHING (submit via PR), VETTING (the safety model), MODES, ARCHITECTURE, ROADMAP.
Related MCP server: Lab Registry Server
Quickstart
npm install
npm run build
npm test
node dist/cli.js list
node dist/cli.js install jolly --dest ./muse-assets # copies files, verifies sha256 before AND after, writes install-receipt.json
node dist/cli.js init my-avatar --kind avatar # scaffold a listing of your own
node dist/cli.js validate my-avatar
node dist/cli.js index # rebuild registry/index.json
node dist/cli.js doctor # registry healthPublish a listing
muse-marketplace init <your-slug> --kind avatarReplace the placeholder SVGs /
persona.mdwith your real assets.Run
muse-marketplace validate <your-slug>until clean, thenmuse-marketplace index.Open a PR. Full walkthrough: docs/PUBLISHING.md.
The vetting / safety model (short version)
Nothing gets to call itself safe here without receipts — the same rule we run across the fleet: anything listed as safe must be validated with real checks.
No executables, ever. Only allowlisted media types (SVG/PNG/JPEG/GIF/WebP, markdown, plain text, JSON, audio). Validation rejects executable extensions outright.
Checksums on everything. Every file in every listing records its sha256 + byte size.
validaterecomputes them;installverifies before and after copying and refuses tampered listings.Path safety. Absolute paths and
..traversal entries are rejected; installs stay inside the destination you name.Vetted is earned.
safety.vettedflips totrueonly after the VETTING.md checklist is run by a reviewer — not by the author checking their own box.Personas are plain markdown you can read in full before using — the file is the payload, nothing hidden.
Full model: docs/VETTING.md · Security reports: SECURITY.md
MCP usage
{ "mcpServers": { "muse-marketplace": { "command": "npx", "args": ["-y", "@reneturcios/muse-marketplace", "mcp"] } } }Tools: marketplace_search · marketplace_list · marketplace_show · marketplace_validate · marketplace_install (install is confined to a caller-provided destination under the current working directory).
Roadmap
Ratings & reviews, signed listings, voice previews in the site player, a publish command that opens the PR for you, federation with other registries, and first-class wiring for Muse's native avatar sharing when it lands. Detail: docs/ROADMAP.md.
MIT. By Rene Turcios / CloudCorpRecords. Fleet: https://github.com/CloudCorpRecords
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Public registry of specs AI agents can build from. Search, pull and publish over MCP.
Search, fetch, lint, and install Agent Skills (SKILL.md) from the SkillMD registry.
Resolve, search, and validate AT Protocol lexicons.
Search, read and install agent-ready React components, screens and blocks for AI products.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceEnables AI agents to discover, install, and manage SKILL.md skills from a Git-backed registry via MCP tools for search, install, and list operations.35 npm2MIT
- FlicenseAqualityBmaintenanceEnables MCP clients to discover and fetch skills, agents, commands, and hooks from the Gen-e2 Lab Registry, and check local inventory compliance against the registry.15-
- AlicenseNot gradedqualityDmaintenanceDiscovers and manages portable agent capabilities (skills and MCP servers) from configurable collections, providing search, inspection, and local installation via CLI and MCP tools.1MIT
- AlicenseNot gradedqualityCmaintenanceEnables agents to install compatible SKILL.md skills from Git repositories or local directories into a persistent registry, then discover them through search, progressive loading, and asset reads. It exposes only user-selected hot skills as individual activation tools, keeping execution and trust decisions under explicit operator control.MIT