Open-source autonomous agents 6 min

Set up an autonomous agent (worked example)

Hermes Agent install — pattern transfers to most others.

Most autonomous agents install the same way: one command, an env var with your API key, optional config. Worked example: Hermes Agent. The pattern transfers to aider, OpenHands, Goose, OpenClaw.

Path A · As a Claude Code plugin (easiest)

/plugin marketplace add sypsyp97/claude-hermes
/plugin install claude-hermes@claude-hermes
/claude-hermes:start

Wizard asks for: model, heartbeat, Telegram/Discord bots, security pin.

Path B · Standalone (curl)

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash
echo "ANTHROPIC_API_KEY=sk-ant-…" >> ~/.hermes/.env
hermes model

What you get

• Persistent. Lives in the background; remembers across sessions.
• Scheduled. "Every Monday at 9, scan my inbox."
• Multi-channel. Terminal, Telegram, Discord, voice.
• Provider-agnostic. Switch between Anthropic, OpenAI, Ollama in one command.

The same shape, other agents

# aider
pip install aider-chat
export ANTHROPIC_API_KEY=sk-ant-…
cd my-project && aider --model sonnet

# OpenHands (Docker)
docker run -p 3000:3000 -e LLM_API_KEY=sk-… \\
  ghcr.io/all-hands-ai/openhands:latest

# OpenClaw
npm install -g openclaw
openclaw init
export ANTHROPIC_API_KEY=sk-ant-…
openclaw run

# Goose
brew install pressly/tap/goose
goose configure
goose session