Projscan
Give your coding agent deep, local repo intelligence across many languages without sending the whole codebase to the cloud.
Overview
ProjScan is an MCP server for the Build phase that exposes 45 local tools so agents can understand multi-language repositories without cloud indexing.
What is this MCP server?
- 45 MCP tools for structure, search, and code intelligence in one local server
- Supports 11 programming languages for polyglot monorepos and side projects
- Runs locally over stdio via npm (projscan mcp) — no remote code upload required
- Agent-first design aimed at repo understanding, not one-off snippet lookup
- Version 4.1.0 with stdio transport for standard MCP client configs
- 11 supported languages
- Server version 4.1.0
Community signal: 4 GitHub stars.
What problem does it solve?
Agents burn context and make mistakes when they lack reliable, structured access to how your repo is actually organized.
Who is it for?
Solo builders running local agents on real codebases who want polyglot repo maps and tool-based exploration in one MCP package.
Skip if: Teams that need a hosted, org-wide code search product with SSO and managed indexes only in the cloud.
What do I get? / Deliverables
After registration, your agent can query repo structure and code intelligence through MCP instead of fragile manual file dumps.
- MCP tool access for repo structure and code intelligence
- Agent workflows that use structured queries instead of full-file pastes
- Local-only analysis path for sensitive codebases
Recommended MCP Servers
Journey fit
ProjScan sits in Build because it powers day-to-day implementation: navigation, structure, and tool-driven analysis while you ship features. Agent-tooling is the canonical shelf for MCP servers that extend Claude Code, Cursor, or Codex with repository-specific capabilities.
How it compares
Local MCP code-intelligence server, not a hosted GitHub Copilot Enterprise or static documentation site.
Common Questions / FAQ
Who is ProjScan for?
Indie developers and small teams using MCP-enabled coding agents who want local, multi-language repository understanding with dozens of structured tools.
When should I use ProjScan?
Use it during Build when onboarding to a repo, planning refactors, or answering architecture questions without uploading the full tree to a remote service.
How do I add ProjScan to my agent?
Add an MCP stdio server entry for the npm package projscan with the required positional argument mcp, then restart Claude Code, Cursor, or your MCP client.