Cluster · Model Context Protocol
MCP servers built for real work
Yoda Digital maintains six Model Context Protocol servers that expose tools, prompts, and resources from production systems: GitLab with 86 tools, Moldova's official ANSC and MTender data, Google Analytics 4 and Search Console, plus end-to-end encrypted messaging between Claude Code instances. Every server is published on npm under a short name and starts with a single command.
Related projects
mcp-gitlab-server
AI DevTools / MCPGitLab MCP server with 86 tools — full GitLab control from any AI agent.
- Stars
- 48
- Forks
- 19
- Version
- v0.3.1
- License
- MIT
npm install -g @yoda.digital/gitlab-mcp-server
→
ansc-mcp-server
Moldova Public ProcurementMCP server exposing Moldova's ANSC public-procurement appeals, decisions, hearings, and PDFs to AI agents.
- Stars
- 0
- Forks
- 0
- Version
- v1.0.3
- License
- MIT
- Tools
- 18
npx -y mcp-ansc-server
→
mtender-mcp-server
Moldova Public ProcurementMCP server for MTender — Moldova's OCDS 1.1.5 public-procurement data, agent-readable.
- Stars
- 0
- Forks
- 0
- Version
- v3.1.1
- License
- ISC
- Tools
- 14
npx -y mtender-mcp-server
→
gacli
SEO & AnalyticsFull-featured Google Analytics 4 CLI + MCP server — production GA4 automation for humans and agents.
- Stars
- 0
- Forks
- 0
- Version
- v1.1.0
- License
- MIT
npm install -g @nalyk/gacli
→
gsccli
SEO & AnalyticsSenior-grade Google Search Console + Indexing API CLI — real production SEO workflows.
- Stars
- 0
- Forks
- 0
- Version
- v1.2.1
- License
- MIT
npm install -g @nalyk/gsccli
→
mycelium
Agent InfrastructureEnd-to-end encrypted messaging between Claude Code instances — secure agent-to-agent coordination.
- Stars
- 0
- Forks
- 0
- Version
- v0.1.1
- License
- MIT
- Tools
- 4
npm install -g @yoda.digital/mycelium
→
Why this matters
MCP has become the lingua franca for plugging AI agents into operational systems. The difference between a demo MCP server and a production-grade one shows in the details: how it handles authentication, how it paginates past API caps, whether prompts are pre-curated for real workflows, whether transports cover stdio plus stateful HTTP.
The six servers here aren't built to tick a technology box — each solves a concrete problem the Yoda Digital team uses in production: DevOps control through agents (mcp-gitlab-server), civic intelligence on Moldova's public procurement (ansc-mcp-server, mtender-mcp-server), full SEO automation (gacli, gsccli), and secure agent-to-agent coordination (mycelium).
Every server declares the supported MCP spec version, exposes a tool list visible in its README, and can be inspected before install via the per-project comparison page and the install page with copy-paste configuration snippets for Claude Desktop, Cursor, Zed, and VS Code.
How to start
- 01 Pick a server from the list below that matches your workflow (DevOps, public procurement, SEO, or agent coordination).
- 02 Open the project's install page — each ships copy-paste configurations for the popular MCP clients.
- 03 Start with a read-only test — every server has either an explicit read-only mode or purely read tools for safe bootstrapping.
- 04 Extend to write tools only after you understand the effects — the README tools section for each project carries the full signature.
The MCP spec evolves quickly (the current version tracked by these servers is 2025-11-25). The portal displays the version exactly as each repo declares it, regenerated on every commit via GitHub webhook.