Run shared-memory agent workspaces across Codex, Claude Code, and MCP with holaOS
Coordinate multiple coding agents in one local-first workspace with shared memory, tools, files, browser access, and MCP-backed integrations.
npx skills add agentskillexchange/skills --skill run-shared-memory-agent-workspaces-across-codex-claude-code-and-mcp-with-holaos
https://github.com/holaboss-ai/holaOS.git, copy apps/desktop/.env.example to apps/desktop/.env, run npm run desktop:install, run npm run desktop:prepare-runtime:local, then start the desktop development workspace with npm run desktop:dev.Use holaOS when an operator needs a durable local workspace where Codex, Claude Code, and the built-in holaOS agent can work over the same project context instead of rebuilding setup in separate tools. The agent workflow is to start a desktop workspace, connect the required files, apps, browser, integrations, and MCP tools, then hand tasks between agents while preserving shared memory and operator-visible state. Invoke this when the job depends on continuity across agents, local files, browser actions, and connected tools; use the normal product UI directly when a human only wants to chat with one assistant or manage the app itself. The scope boundary is the repeatable multi-agent workspace operation: shared memory, tool routing, and cross-agent handoff. It is not a generic desktop app listing, model SDK listing, or broad MCP catalog entry.
What this skill actually does
Inputs and prerequisites: holaOS desktop workspace, Node.js 24 and npm for OSS development setup, optional Codex or Claude Code agent, optional MCP integrations.
Setup notes: Clone https://github.com/holaboss-ai/holaOS.git , copy apps/desktop/.env.example to apps/desktop/.env , run npm run desktop:install , run npm run desktop:prepare-runtime:local , then start the desktop development workspace with npm run desktop:dev .
Source and verification boundary: use https://www.holaos.ai/docs/getting-started as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.