Build With Iris.Md.

Install the Iris.Md agent on Linux, macOS, or WSL with a single shell command.

Boot the agent, connect a model, and have your first conversation.

Configure models, channels, sandboxes, and memory through ~/.iris/config.toml.

Bring your own keys for OpenAI, Anthropic, Google, or any local OpenAI-compatible endpoint.

Spawn isolated subagents with their own context, terminal, and lifecycle.

Long-term memory that survives restarts and gets richer the longer the agent runs.

Auto-generated, reusable procedures the agent writes for itself.

Five backends: local, Docker, SSH, Singularity, and Modal.

Talk to your agent from any Telegram chat — DMs, groups, or channels.

Run the agent as a Discord bot with slash-command support.

Mention the agent in any Slack channel or DM it directly.

Connect via the WhatsApp Cloud API for end-to-end agent messaging.

Send and receive over IMAP/SMTP — perfect for scheduled briefings.

The terminal is a first-class channel.

Open PRs, triage issues, and review diffs straight from the agent.

Create, update, and triage Linear issues from any conversation.

Read and write Notion pages and databases as long-term shared memory.

Gmail, Calendar, Drive, and Docs through a single OAuth connection.

Post, reply, and monitor mentions on X via the v2 API.

Federated, end-to-end encrypted chat for self-hosted deployments.

Private, end-to-end encrypted messaging via signal-cli.

Generic inbound/outbound webhooks for any service not listed above.

Full reference for the iris command-line interface.

Every key accepted by ~/.iris/config.toml.

HTTP API exposed by the local Iris.md.

Extend the agent with custom tools written in Python or TypeScript.