Skip to main content
Glama
David-Sat

skill-hub

by David-Sat
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Local

(ノ^_^)ノ┻━┻ SkillDesk

Unified Skill & MCP Manager for AI Coding Agents.

License: MIT MCP Privacy

NOTE

Security & Privacy Guarantee: skilldesk runs 100% locally on 127.0.0.1 and communicates with coding agents via standard stdio (MCP). There is zero telemetry, no tracking pixels, and no remote analytics. Registry checks are unauthenticated read-only calls directly to public NPM, PyPI, and GitHub APIs. Audited in PRIVACY.md and SECURITY.md.

SkillDesk scans all installed coding agent harnesses across your machine—showing what skills and MCP servers are active, profiling their startup context window token footprint, inspecting bundled helper scripts, safely removing unused skills with automated backups, and letting you 1-click update outdated MCP packages.

Quickstart

Method

Command

Purpose

Instant Dashboard

npx skilldesk

Launch local web dashboard at http://127.0.0.1:4350

Terminal Audit

npx skilldesk audit

Fast terminal ASCII overview of active harnesses & tokens

In Agent

/skills

In-chat audit & dashboard trigger

Global Setup

npm i -g skilldesk && skilldesk init -g

Interactive setup for Antigravity, Claude Code, Cursor, Windsurf

(Both skilldesk and skill-desk command invocations are supported)

Related MCP server: OSource-Manager

Core Capabilities

  • Cross-Harness Discovery: Detects active skills and MCP servers across Antigravity, Claude Code, Cursor, and Windsurf in one unified view.

  • Turn-0 Context Profiler: Measures startup token overhead and flags heavyweight skill descriptions before they bloat agent context.

  • Safe Skill Removal: Removes unwanted skills with automated, timestamped safety backups.

  • 1-Click MCP Updates: Scans NPM, PyPI, and GitHub registries to upgrade outdated server configs.

  • In-Agent Tooling: Native MCP tools and in-chat slash commands directly inside your AI coding sessions.

System Architecture

┌─────────────────────────────────────────────────────────────┐
│                         SKILLDESK                           │
├──────────────────────────────┬──────────────────────────────┤
│ CLI / AGENT ENTRY            │ EMBEDDED PREACT DASHBOARD    │
│ • Terminal: npx skilldesk    │ • 127.0.0.1:4350 (28 KB JS)  │
│ • In-Agent: /skills          │ • Token gauges & File tree   │
│ • MCP stdio: skilldesk mcp   │ • 1-Click Update triggers    │
├──────────────────────────────┴──────────────────────────────┤
│ LOCAL CORE ENGINE (Node / TypeScript)                       │
│ ├── scanner:  Cross-platform harness & config discovery     │
│ ├── profiler: BPE token weighting & asset classification    │
│ ├── updater:  Concurrent NPM/PyPI/GitHub version checks     │
│ └── server:   Zero-dependency local HTTP server             │
└─────────────────────────────────────────────────────────────┘

License

MIT © David Satomi

This server cannot be deployed

Shareable, User-Owned Context BackendPeople, applications, and agents read, write, and coordinate together in one context lake. Reach it over MCP, a CLI, a Python SDK, or REST.fulcradynamics.comAd

Maintenance

ActivityMaintained
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • KeyStack

    nzt108-dev
    F
    license
    Not graded
    quality
    C
    maintenance
    Local dashboard and MCP server that maintains a live project registry and skill library, enabling coding agents to pull project context and automatically update status, stack, and next steps.
    -

  • skillcp

    lum1n
    A
    license
    Not graded
    quality
    B
    maintenance
    Enables managing a canonical library of agent skills and MCP servers, syncing them across multiple development harnesses, and adding, importing, or configuring them through MCP tools.
    6 npm
    MIT

  • SkillDock

    BK927
    A
    license
    Not graded
    quality
    B
    maintenance
    Enables 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