Aiden CLI
A Node.js terminal agent that takes a goal, works with files, shell and browser and backs its result with evidence
- 882 ★ on the repository
- Last commit
- AGPL-3.0
Broad access
Why this level
- The agent runs shell commands, drives a browser and can post into channels, and daemon mode starts work without a person.
Install
- Services
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/aidenDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The agent runs shell commands, drives a browser and can post into channels, and daemon mode starts work without a person.
This is third-party code. Review the repository files before installing.
What it does
Aiden starts with the `aiden` command and opens an operator terminal interface. You give it a goal and it plans and carries out the steps: it reads and edits files, runs commands, visits websites through browser automation and calls tools over MCP. Every action is recorded as an attempt and a result, and completion is verified separately: a model saying it is done does not count as proof. In codebase mode Aiden snapshots the repository, makes changes that do not collide with your edits, runs tests and builds and keeps their results in a log. For reviewing runs there are `aiden runs list` and `aiden runs show <id>`. Daemon mode is opt-in and starts work on events: a file change, a webhook, an email or a schedule. The model is chosen from Anthropic, OpenAI, Groq, Gemini, OpenRouter, DeepSeek, Mistral, Kimi, Ollama and any OpenAI-compatible endpoint. Replies are also available in Discord, Slack, Telegram, Email and other channels. Bundled skills with `SKILL.md` install with `/skills install <name>`, and the MCP client works over stdio and Streamable HTTP.
Who it is for. Developers and administrators who want an autonomous terminal agent with a verification log, their own model keys and support for Windows, Linux or WSL.
Good fit when
- Tests fail and the agent has to find the cause, fix the code, rerun the check and show what changed
- A repository is unfamiliar and you want a breakdown of its structure based on files, without model guesses
- A task has to run without a person: on a file change, a webhook or a schedule through daemon mode
- You want the agent's answers in Slack, Telegram or Discord, not only in the terminal
- You want to switch models freely between a cloud provider and local Ollama without changing the tool
Not a fit when
- You need a plugin or skill inside the Claude Code or Cursor you already use: Aiden installs as a separate agent
- The main system is macOS and you need the full feature set: the author marks macOS as API mode
- You do not want to give an agent shell and browser access: useful work depends on those permissions
- You need a ready-made Windows installer: the author deferred it and publishes only the npm package
Example request
Find the cause of the failing tests, fix the root problem, rerun the relevant checks and show me the changes.Limitations
Aiden runs commands and drives a browser, so it is better to do the first runs in a separate folder or a Docker sandbox. The system classifies dangerous commands and asks for confirmation, but the trust level is set by the user.
macOS support is limited to API mode. The author deferred the Windows installer and publishes an npm package that needs Node.js 20 or 22. The core is under AGPL-3.0 and the skills are under Apache-2.0.
You bring your own model keys, and the package pays for nothing. The last commit in the repository is dated September 2026 and versions ship quickly, so read the release notes before updating.
How to disable. Stop the daemon (check it with `aiden daemon status`) and remove the service, then run `npm uninstall -g aiden-runtime`. Delete the settings and keys directory by hand.
FAQ
How do I install Aiden?
Run `npm install -g aiden-runtime`, then start `aiden`. For a run without installing there is `npx aiden-runtime@latest`, and `aiden web` opens the workbench in a browser.
Which models are supported?
Anthropic, OpenAI, Groq, Gemini, OpenRouter, DeepSeek, Mistral, Kimi, Ollama and any OpenAI-compatible endpoint. You enter the keys yourself during the first setup.
How do I turn on work without a person?
Daemon mode is off by default. Run `aiden daemon start` and add a trigger, for example `aiden trigger add cron --schedule "<expression>"`.
How are skills installed?
With `/skills install <name>` in the agent chat. `/skills` lists the installed ones, and the community registry is at skills.taracod.com.
Does Aiden work on macOS?
Partly. The author calls macOS API mode and fully supports Windows, Linux and WSL.
Discussion
No questions or reviews yet.
Related
All Telegram toolsA CLI to run several coding agents: one config, a session console, local memory and a set of engineering skills
A compact OpenClaw alternative: a personal assistant on Claude Agent SDK with agents in isolated containers and messengers added via skills
A local daemon assembles Claude Code, Codex and OpenCode into a team: tasks, separate worktrees, checks and merging.
Open-source personal AI assistant on your own machine: answers in Telegram, Slack, Discord and WhatsApp, extended with skills and plugins
Claude-to-IM
claude-to-im
A bridge between Claude Code or Codex and messengers: talk to the agent from Telegram, Discord, Feishu, QQ or WeChat
A Claude Code plugin that gives the agent a name, a personality and memory across sessions, plus chat channels.
README in short
The project calls itself an autonomous work engine: it plans, acts, recovers after failures and proves the result. The current stable version 4.21.2 ships as the `aiden-runtime` package. The documentation details versions 4.5 to 4.21, the CLI command list and installation on Windows, Linux and WSL.