Getting Started

This guide takes you from zero to a running Dryade install, a configured model provider, and your first agent — in about ten minutes on a fresh machine.

---

1. Install Dryade

Download the native installer for your operating system from dryade.ai/download:

OS	File	Notes
Windows 11+	Dryade-<version>.exe	Authenticode-signed
macOS (Apple Silicon)	Dryade-<version>.pkg	Developer ID + notarized
Linux (glibc-2.35+)	dryade_<version>_amd64.deb	Tested on Ubuntu 24.04

The installer is single-binary. It bundles the platform, the workbench UI, the plugin manager, the embedded inference runtime, and a local KV/vector store. No Postgres, Redis, or Qdrant to set up separately.

Launch the app after install. On first run, Dryade auto-generates a free-tier configuration and opens the workbench at http://localhost:3000.

To unlock team or enterprise features later, drop a license.dryadelicense file anywhere on disk — Dryade picks it up on the next start.

Sovereign / Kubernetes deployments: the Docker / Helm / Compose path is documented separately at Deployment. It targets operators who already manage their own runtime dependencies (Postgres or pgvector, an inference endpoint, observability). The bundled-dependencies posture above is the native-installer surface only.

---

2. Configure a Model Provider

Open the workbench, go to Settings → Model Providers, and pick one of:

• Local model (default) — the embedded inference runtime starts automatically with the recommended free-tier model. No API key needed.
• OpenAI / Anthropic / Mistral / OpenRouter / Together AI — paste your API key. The workbench validates it with a one-token completion before saving.
• Self-hosted vLLM / Ollama / TGI — set the base URL of your inference endpoint. The workbench tests the /v1/models route to confirm reachability.

You can mix multiple providers and route different agents to different models. See Workflows for the routing rules.

---

3. Install a Plugin

The plugin catalog lives at dryade.ai/marketplace. Browse, pick a plugin, click Install — the workbench downloads the signed .dryadepkg, verifies the signature against the marketplace allowlist, and registers the plugin’s agents, tools, and routes with your local platform.

Default tier (free) lets you install community-licensed plugins. Team and enterprise tier plugins require the matching license.

---

4. Create Your First Agent

From the workbench:

1. Agents → New Agent
2. Pick a base template (Chat, ReAct, Plan-then-Execute, or one of the templates a plugin contributed)
3. Wire in the tools and models you want the agent to use
4. Save & Test — the workbench opens a chat panel running against the agent in isolation

Once you are happy with the agent, expose it via a workflow (programmatic API), a chat surface (user-facing), or both. See Workflows for the orchestration layer.

---

5. Use the REST API (optional)

For programmatic access, every workbench action is also reachable via the Dryade REST API.

Base URL (native installer): http://localhost:8000/v1 Base URL (cloud / sovereign deployments): the URL of your Dryade host.

Authentication

Generate a personal access token under Settings → API Keys, then:

export DRYADE_API_KEY="dak_..."
curl -H "Authorization: Bearer $DRYADE_API_KEY" \
  http://localhost:8000/v1/agents

import requests
response = requests.get(
    "http://localhost:8000/v1/agents",
    headers={"Authorization": "Bearer dak_..."},
)

const response = await fetch("http://localhost:8000/v1/agents", {
  headers: { Authorization: "Bearer dak_..." },
});

Full reference: API Reference.

---

Next Steps

• Plugin Development — build and publish your own plugin
• Workflows — chain agents and tools into multi-step pipelines
• Security — security model, allowlist, license lifecycle
• Deployment — sovereign, cloud, Kubernetes deployments