Start Here · 01.04
Bootstrap Your First Agent
Wire a durable Agentis identity to a model runtime with one CLI command, agentis bootstrap, then verify it's alive from the dashboard.
What bootstrapping actually does
Bootstrapping is not "starting a chat." It's the one-time act of wiring a durable Agentis identity to a model runtime — creating the agent record that will go on accumulating memory and skills for as long as you keep it, and pointing it at whichever CLI harness or HTTP endpoint should execute its turns today. That binding is exactly the thing you can change later without losing anything, via the Runtime Abstraction Layer.
The bootstrap command
agentis bootstrap \
--url http://127.0.0.1:3737 \
--api-key $AGENTIS_API_KEY \
--name "The Brain" \
--adapter codex
pnpm exec tsx packages/cli/src/index.ts bootstrap \
--url http://127.0.0.1:3737 --api-key $AGENTIS_API_KEY \
--name "The Brain" --adapter codex
Choosing an adapter
| Adapter | Runtime |
|---|---|
claude_code | Claude Code CLI |
codex | OpenAI Codex CLI |
cursor | Cursor |
hermes_agent | Hermes Agent (ACP) |
openclaw | OpenClaw |
http | Any HMAC-authenticated custom endpoint |
Don't agonize over this choice — it's not permanent. Bootstrapping just picks who executes today's turns; the identity, memory, and skills that accumulate from here belong to Agentis regardless of which adapter is attached.
Confirm it worked
Open the dashboard's Agents view. A successful bootstrap shows your new agent with a live status and its bound adapter. From here you have a real, durable, empty-minded agent — the next move is deciding whether to start it from zero or bring in an agent you already run elsewhere, memories and skills included.
Continue
Clone the repo, run pnpm install, and start the Agentis API on 127.0.0.1:3737 — then open the dashboard and mint your first API key.
Describe intent, build, validate, dry-run, run, and deliver — the paved road every Agentis workflow follows, walked end to end with a real example.