vault
A minimal, keyboard-first file manager that replaces Finder and Explorer. Search-first navigation, instant preview, OS-trash-only deletes. Go + Wails.
Install / Use
claude mcp add arxdsilva -- npx -y github:arxdsilva/vaultIf the server publishes to npm under a different name, use that package instead — check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
Development & EngineeringSupported Platforms
Our assessment of vault
vault scores 81/100 on our quality scale, 274th of 1,340 Development & Engineering skills we index (top 21%).
Its MCP Server is 7.3 KB long, well organised into 17 sections with 8 code examples: a thorough specification that gives an agent plenty to work with.
It has 3 GitHub stars, so there is little community track record yet; judge it on its content.
Maintenance, license and trust
- The repository was last updated 14 days ago, so vault is actively maintained.
- Our last check on 2026-09-02 found the source still online.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 92/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands.
Automated pattern scan on 2026-09-24. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
vault compared with similar skills
All 4 of these similar skills score higher than vault; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| vault (this skill)by arxdsilva | 81 | 3 | 14d ago | MCP Server |
| Agent-Reachby Panniantong | 100 | 85.2k | 8d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 73.7k | today | CLAUDE.md |
| rufloby ruvnet | 100 | 73.2k | today | CLAUDE.md |
| CowAgentby zhayujie | 100 | 47.1k | today | CLAUDE.md |
Frequently asked questions
- How do I install vault?
- Run
claude mcp add arxdsilva -- npx -y github:arxdsilva/vault. The install tabs above show the steps for each supported agent. - Which AI agents does vault work with?
- It is written for Claude Code and Claude Desktop, as a MCP Server file. Other agents that read the same format can often use it too.
- Is vault safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. It is MIT-licensed and scores 92/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is vault still maintained?
- The repository was last updated 14 days ago, so vault is actively maintained.
Skill content
View source on GitHubVault
A minimal, cross-platform file manager aiming to replace Finder (macOS) and Windows File Explorer — fast, lightweight, and free of feature bloat.
A file manager that behaves like a command palette, not a filing cabinet: search-first navigation,
zero chrome, keyboard-first, instant everything. Full design spec: docs/SPEC.md.

Goals
- Fast startup, low resource usage
- Keyboard-first navigation, full mouse support
- Cross-platform: macOS and Windows at minimum (Linux as a stretch goal)
- Clean, minimal UI — no unnecessary chrome or feature creep
- Feels native on each platform it targets
Stack
- Shell: Wails v3 (Go backend + native webview)
- Frontend: React 18 + TypeScript + Tailwind CSS, virtualized lists/grid via TanStack Virtual, Zustand for UI state, Shiki for code preview syntax highlighting
- Search: Go-side fuzzy matcher (
sahilm/fuzzy) over an in-memory index of the home directory, ranked by match quality → recency → frequency; incremental updates viafsnotify - Trash: OS-native trash/recycle-bin per platform adapter — never a permanent delete
- Thumbnails: image thumbnails decoded/resized/cached on disk by a bounded Go worker pool
Architecture
main.go Wails app entry: window, theming, service registration
cmd/vault-mcp/ MCP server entry point (Claude Desktop / Cursor / MCP clients)
internal/core/ platform-neutral: dir listing, file ops, search index, frecency,
fsnotify watchers, thumbnail generation — no OS-specific APIs
internal/platform/ NewTrasher() build-tag dispatch shared by main.go and cmd/vault-mcp
internal/platform/darwin/ macOS adapter: Trash (~/.Trash)
internal/platform/windows/ Windows adapter: Recycle Bin (SHFileOperationW) — cross-compiles
clean but is unverified at runtime (no Windows box in dev/CI here)
internal/service/ thin Wails-bound services (Files, Search, Pinned) exposed to the
frontend; call only into internal/core
internal/mcp/ MCP tool server (9 tools) wrapping internal/core for cmd/vault-mcp
frontend/ React + TypeScript + Tailwind UI; talks to internal/service via
generated Wails bindings (frontend/bindings/); lib/chatTools.ts +
lib/llmClient.ts + components/ChatPanel.tsx are the embedded
AI assistant, calling the same lib/vault.ts operations
See .cursor/rules/architecture.mdc for the full boundary
rules (core stays UI-agnostic; OS-specific behavior stays in one adapter per OS).
Keyboard map

| Key | Action |
|-----|--------|
| Type anywhere | Search (top bar is always focused on launch) |
| Esc | Clear search, return to Home |
| ↑ ↓ | Move selection in a list/search results |
| ← → ↑ ↓ | Move selection in grid view |
| Enter | Open selection |
| Cmd/Ctrl + Enter (in search) | Reveal result's folder in-app |
| Cmd/Ctrl + ↑ | Go to parent folder |
| Space | Quick preview (image/video/audio/PDF/text/code) |
| ← → ↑ ↓ (preview open) | Move to previous/next file |
| Cmd/Ctrl + K | Command palette |
| F2 | Rename selection |
| Cmd/Ctrl + N | New folder |
| Cmd/Ctrl + C (via palette "Copy") / Cmd/Ctrl + V | Copy / paste |
| Cmd/Ctrl + Z | Undo (rename/move only) |
| Drag file/folder onto the pinned rail | Pin it; drag within the rail to reorder |
Non-goals (MVP)
No tabs, split panes, cloud integrations, network drives, bulk metadata editing, a general
third-party plugin system, or file tagging. See docs/SPEC.md §7 for the full list — this does
not exclude the AI integration described below, which is a defined, first-party feature.
AI Integration
Vault exposes the same 9 file operations (list directory, search, read file, move, rename,
paste-into, new folder, trash, list pinned) two ways — see docs/SPEC.md §10 for the design
rationale:
MCP server (Claude Desktop, Cursor, and other MCP clients)
make mcp-build # builds bin/vault-mcp
./bin/vault-mcp --root ~/Projects # optional: restrict to a directory (defaults to $HOME)
./bin/vault-mcp --http :8090 # optional: serve over streamable HTTP instead of stdio
Add it to Claude Desktop's claude_desktop_config.json:
{
"mcpServers": {
"vault": {
"command": "/absolute/path/to/bin/vault-mcp"
}
}
}
Or to Cursor's .cursor/mcp.json (project-level) or ~/.cursor/mcp.json (global):
{
"mcpServers": {
"vault": {
"command": "/absolute/path/to/bin/vault-mcp"
}
}
}
File operations are confined to --root (default $HOME) — a path outside it is rejected.
Deletion always goes to the OS trash, never a permanent delete.
Embedded chat panel
Click the assistant icon in the top bar, then set a provider (Claude or ChatGPT) and API key in
Settings. The key is stored locally in the browser's localStorage — see docs/STATUS.md for the
plaintext-storage caveat.
Performance
Measured against the four targets in docs/SPEC.md §8 — cold start, search latency at 200k
files, scroll at 50k entries, idle RAM — on an Apple M3 Pro, production build:
Three of four comfortably clear their target; search latency mostly does, with one worst-case
gap (a broad query, scoped to a folder, matching every entry) — see
docs/PERFORMANCE.md for the full numbers, the caveat, and exactly how
each figure was produced (all reproducible: a committed Go benchmark plus three manual
procedures).
Status
All ten build-order steps (including AI integration) are implemented and validated. See
docs/STATUS.md for the checklist and a list of known limitations (Windows
runtime verification, video thumbnails, "Open With…", .msi packaging, chat API key storage).
Development
- Design spec and non-goals:
docs/SPEC.md - Coding conventions and workflow live in
.cursor/rules/(TDD, core/UI/platform-adapter boundaries, commit discipline, response style) - Commits are prepared by AI tooling but always reviewed and executed by a human
(
.cursor/rules/git-commits.mdc)
Setup
go install github.com/wailsapp/wails/v3/cmd/wails3@v3.0.0-alpha2.117 # once, if wails3 isn't on PATH
cd frontend && npm install && cd ..
Run
wails3 dev # dev mode, hot reload
wails3 build && ./bin/vault # production build, run directly
Package
wails3 package # bin/vault.app (ad-hoc signed)
wails3 task package:dmg # bin/vault.dmg (macOS only, via hdiutil)
wails3 task windows:package GOOS=windows # NSIS .exe installer (cross-compiled; see docs/STATUS.md)
Test
go build ./...
go vet ./...
go test -race -cover ./... # internal/core, internal/mcp, internal/service, internal/platform/darwin
GOOS=windows go build ./... && GOOS=windows go vet ./... # windows adapter compile check
cd frontend
npx tsc --noEmit
npx vitest run
npm run build
golangci-lint isn't installed in this environment; run golangci-lint run if you have it
available locally.
License
Related Skills
Agent-Reach
85.2kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
headroom
73.7kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
ruflo
73.2k🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
CowAgent
47.1kOpen-source super AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, multi-channel. Lightweight, extensible, one-line install.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
