7 days

Use the first week as a practical Hermes tutorial route.

This route is deliberately sequential. The goal is not to read everything; the goal is to leave the first week with a Hermes setup you can actually operate.

Built around Hermes v0.9Updated: 2026-04-18Community guide

One concrete outcome for each day

Sequenced to match how a real Hermes setup matures

Designed to keep beginners moving instead of collecting unfinished tabs

Day 1

Install Hermes and confirm the first reply

You can run `hermes --version`, open the CLI, and get a real response back.

Return to setup

Tasks

  • Use the setup guide for your operating system and complete the official installer flow.
  • Reload the shell and confirm the CLI exists before you touch provider settings.
  • Run `hermes setup`, then send the first short prompt to make sure the tool is alive.
  • Try `hermes --fast` once so you know what the direct execution path feels like.

Watch for

  • `hermes` not found after install means your PATH still needs attention.
  • A silent setup flow usually points to missing provider credentials or shell state not being reloaded.

Day 2

Lock in provider configuration

Hermes answers reliably with the provider and model you expect.

Tasks

  • Check which model the config currently points at and update it if you need a different default.
  • Store the provider key using the current Hermes auth flow or the config path you trust.
  • Test a small reasoning prompt and one simple command-like request so you cover both normal and practical use.

Watch for

  • Provider errors that mention auth or rate limits are easier to fix now than later in the week.
  • If the model returns empty responses, confirm both the key and base URL for the provider you chose.

Day 3

Connect Hermes to a real chat surface

A channel or gateway route sends a real message through Hermes and returns a reply.

Tasks

  • Pick one delivery surface such as Telegram, Feishu, or another supported channel.
  • Run the gateway setup for that platform and keep token handling explicit.
  • Send a test message from the real channel instead of relying on local assumptions.
  • Check whether the dashboard or gateway logs tell the same story as the user-facing message flow.

Watch for

  • Channel issues often come from expired tokens, outdated webhooks, or another process still holding the same bot identity.
  • A gateway that starts cleanly but never replies still needs end-to-end validation from the actual chat app.

Day 4

Use Hermes as a practical tool, not just a chat shell

Hermes successfully calls one or more tools in a way you can observe and verify.

Tasks

  • Ask for something with external context, such as a weather lookup or current information search.
  • Try a safe terminal-oriented task so you understand where approvals or tool restrictions appear.
  • Notice which tools trigger automatically and which ones need you to phrase the request more clearly.

Watch for

  • If tools never trigger, inspect the enabled toolsets or provider-specific capability settings.
  • If browser or web tools timeout, treat that as an environment or network problem, not as a content problem.

Day 5

Add one skill or extension

Hermes gains at least one new capability that you can trigger intentionally.

Tasks

  • Search the skill ecosystem for one practical extension you can test in minutes.
  • Install it, list installed skills, and trigger it with a phrase that matches its documented entry points.
  • Decide whether the added capability belongs in your everyday workflow or just in experiments.

Watch for

  • A skill that installs but never triggers usually has a naming or invocation mismatch.
  • If a skill conflicts with built-in tools, temporarily narrow the tool surface while testing it.

Day 6

Practice background work and backups

You understand how Hermes behaves when work outlives one foreground shell session.

Tasks

  • Run or monitor one background-style task so you can observe long-lived behavior.
  • Inspect the session or status commands Hermes exposes for process visibility.
  • Create a backup artifact and check how a restore or dry run would work before you actually need it.

Watch for

  • A backup you have never inspected is only a hope, not a recovery plan.
  • Background process monitoring is only useful when you know what output signal should count as success.

Day 7

Debug the workflow and prepare for migration

You can reason about problems, read Hermes state, and move confidently into OpenClaw migration.

Open migration guide

Tasks

  • Use the debug and config views to understand how Hermes reports its state.
  • Write a short personal checklist for the setup you want to keep stable.
  • Read the migration guide and decide whether the current machine is ready for `hermes claw migrate`.

Watch for

  • If you cannot explain where provider, channel, and backup state live, you are not ready to migrate anything important.
  • Treat migration as the last step of understanding, not as the first step of discovery.