Agentlas OS
A layer over coding agents that assembles a temporary specialist team per task inside Claude Code, Codex, Cursor and Gemini
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- The install script edits plugin and MCP config in the agent hosts
- Orchestrates several agents that run tools
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/agentlas-osDetects the agents on your machine, checks the risk and pins the version.
Other ways to install
Run one by one in the Claude Code chat
/plugin marketplace add agentlas-ai/agentlas-os
/plugin install hephaestus@agentlas-core-engineInstall with one command: curl -fsSL https://raw.githubusercontent.com/agentlas-ai/Agentlas-OS/main/scripts/install-all-runtimes.sh | bash. Read the script beforehand, and after install run hephaestus doctor.
Other ways from the author
curl -fsSL https://raw.githubusercontent.com/agentlas-ai/Agentlas-OS/main/scripts/install-all-runtimes.sh | bashThe script verifies the release checksum and writes plugins and MCP into the found hosts; run hephaestus doctor afterward.
This is third-party code. Review the repository files before installing.
What it does
Agentlas OS installs into the agent hosts you already use and adds orchestration: for a given task it selects and assembles a temporary team of specialist agents, for example frontend and backend, and disbands it afterward. Specialists live in a hub, and a separate orchestrator spins up per task. It runs locally and with any model, and installs with a single script that verifies the release checksum and writes plugin and MCP config into the hosts it finds. It bundles a large set of skills and plugins.
Who it is for. For developers who want to add agent-team orchestration to their usual tools.
Good fit when
- You want a specialist agent team assembled per task
- You need orchestration on top of Claude Code, Codex, Cursor or Gemini
- You want a local approach not tied to a single model
Not a fit when
- A single agent without team orchestration is enough
- You do not want to run an install script that edits host configuration
Example request
Assemble a frontend and backend specialist team for this task and split up the workLimitations
Installation runs a script that writes plugin and MCP config into the agent hosts it finds, which is worth understanding before running. Some features tie into the Agentlas cloud and desktop. The orchestration overhead is not worth it for every task.
How to disable. Remove the ~/.agentlas directory and the wrappers from ~/.local/bin, and undo the plugin and MCP entries written into the host configs.
Security check
- The install script edits plugin and MCP config in the agent hosts
- Orchestrates several agents that run tools
README in short
The README presents Agentlas OS as a layer over coding agents that assembles a temporary specialist team per task and disbands it afterward. Specialists live in a hub, a separate orchestrator spins up per task, all local and with any model. Installation is a single script that verifies the release checksum, installs the runtime into the home folder and writes plugins and MCP into the found hosts. Apache-2.0 licensed.
FAQ
Which hosts does it work in?
Inside Claude Code, Codex, Cursor, Gemini and others; the installer finds supported hosts on the machine.
Where does the installer write?
It puts the runtime in ~/.agentlas, wrappers in ~/.local/bin and plugin and MCP config into the found hosts, nothing outside the home folder.
Related
Open-source personal AI assistant on your own machine: answers in Telegram, Slack, Discord and WhatsApp, extended with skills and plugins
A self-improving agent from Nous Research with a TUI, messaging gateway, cron jobs and skills it writes itself
An open source coding agent for the terminal and desktop with build and plan modes
Open prompt library with a Claude Code plugin, MCP server and CLI: search, fetch and improve prompts and skills from an agent