Agent Workspace MCP
Give coding agents a sandboxed workspace with filesystem, bash, and uv-powered Python execution without touching your host machine blindly.
Overview
Agent Workspace MCP is a MCP server for the Build phase that provides a sandboxed filesystem, bash, and uv-powered Python execution for coding agents.
What is this MCP server?
- Sandboxed agentic workspace with filesystem, bash, and Python execution
- uv-powered Python runs inside the isolated environment
- OCI image ghcr.io/hrrodan/agent-workspace-mcp:1.4.9 for container deploy
- stdio transport via published package registry entry
- Agent Workspace MCP branding with SVG logo in server metadata
- Server version 1.4.9
- OCI package ghcr.io/hrrodan/agent-workspace-mcp:1.4.9
- stdio transport documented in server.json
What problem does it solve?
Letting agents run shell and Python on your laptop without isolation is risky and hard to reproduce across machines.
Who is it for?
Solo builders who want reproducible, sandboxed command and Python execution for agent coding workflows.
Skip if: Teams that only need read-only repo search with no execution, or who require fully managed cloud dev environments without self-hosted containers.
What do I get? / Deliverables
After you install the OCI image or stdio server, your agent executes file and command work inside a bounded workspace via MCP tools.
- Isolated filesystem and shell operations callable by the agent
- uv-managed Python execution inside the workspace
- Repeatable containerized MCP server for team or CI agent stacks
Recommended MCP Servers
Journey fit
Canonical shelf is Build because the server is core agent infrastructure used while implementing and running code under guardrails. Agent-tooling subphase reflects secure execution surfaces that extend what Claude Code-style agents can do locally or in containers.
How it compares
Sandboxed execution MCP for agents, not a clinical API server or marketplace directory.
Common Questions / FAQ
Who is Agent Workspace MCP for?
Indie developers and agent authors who need filesystem, bash, and Python tools in a sandboxed MCP workspace.
When should I use Agent Workspace MCP?
Use it while building agent features, evals, or automation that must run commands and scripts with isolation instead of on your bare OS.
How do I add Agent Workspace MCP to my agent?
Pull ghcr.io/hrrodan/agent-workspace-mcp:1.4.9 or configure the stdio package in your MCP client per the GitHub repository instructions.