# agenticworkspace [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/RudrenduPaul/AgenticWorkspace  
**GitHub Stars:** 0  
**npm Downloads (last month):** 221  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/agenticworkspace

## Description
Wraps the AgenticWorkspace CLI as a single generic MCP tool for repo workspace scans.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "agenticworkspace": {
    "command": "npx",
    "args": ["-y","agenticworkspace-cli"]
  }
}
```

## Documentation

## What agenticworkspace MCP server does

The agenticworkspace MCP server connects an AI agent to a local repository and the AgenticWorkspace scan/scaffold pipeline. Its purpose is to turn repository metadata into a structured `.workspace/` directory that agents can use for progressive context and session handoffs.

The underlying project is available as both JavaScript/TypeScript and Python packages. The JavaScript package requires Node.js 20 or newer, while the Python package provides an equivalent CLI and library pipeline. Both implementations use the same `.workspace/` directory shape.

A repository scan can identify JavaScript or TypeScript, Python, and lighter-weight signals for Rust, Go, and Ruby. It also reports package-manager information for npm, pnpm, and Yarn, along with whether the repository is a monorepo and how many packages it contains.

## How it works

The agenticworkspace MCP server delegates repository workspace operations to the AgenticWorkspace CLI rather than implementing a separate repository model. The CLI accepts a target path, scans manifest and configuration files, and writes generated workspace data under the repository's `.workspace/` directory.

The generated structure can include a workspace metadata file, root context markdown, a handoff directory, and adapter files. The documented initialization flow also installs Claude Code hooks for session start, pre-tool calls, and session-end handoff generation. These hooks are configured through files under `.workspace/adapters/claude-code/`.

Existing agent-related configuration is detected without being overwritten. The scan checks for `CLAUDE.md`, `AGENTS.md`, Cursor rules, and GitHub Copilot instructions. It also detects Serena, GitNexus-style configuration, and repo-harness state without reading from or modifying those systems.

## Tools and capabilities

The project documents CLI operations for initialization, scanning, status checks, adapter installation, and creating handoffs. Each subcommand supports structured JSON output, including error cases, so an MCP caller can consume results without parsing terminal-oriented text.

Key capabilities include:

- Detecting repository language, package manager, and monorepo package count.
- Creating `.workspace/` context and handoff files.
- Installing the documented Claude Code adapter and its hook scripts.
- Reporting existing configuration and detected memory backends.
- Identifying interrupted or malformed initialization state instead of silently replacing it.
- Supporting extension through separate memory-backend and adapter interfaces.

Generated shell values pass through allowlist and quoting checks before being embedded in hooks. The project also supports partial-state recovery through an interactive repair, reset, or abort flow, with structured errors in JSON mode.

## Setup and configuration

The repository provides two installation paths. JavaScript/TypeScript users can run `npx agenticworkspace-cli init`, install the package globally, or build it from source with npm. Python users can install `agenticworkspace-cli` from PyPI and run `agenticworkspace init --path /path/to/repository`. The source-based Python workflow installs from the repository's `python` directory.

The documented initialization command accepts a repository path and can emit JSON results with `--json`. A successful run reports the target, detected stack, existing configuration, memory-backend findings, context budget, adapter status, and workspace location.

No MCP-specific environment variables, client configuration, transport, or server launch command is specified in the provided material. Configure those details only from additional project documentation or the package's actual MCP entry point.

## Limitations and notes

This is identified as a version 0.1 release, so its documented scope is limited to the implemented scan, scaffold, and adapter pipeline. The README does not establish support for adapters beyond the working Claude Code adapter, nor does it name specific MCP clients.

The `agenticworkspace` version reported in JSON is an internal schema marker and may differ from the installed npm or PyPI package version. The generated workspace is repository-local, and the tool is designed to avoid modifying detected external memory or agent-configuration systems.

_Full upstream README: https://allmcps.com/mcp/agenticworkspace/readme_