FAQ

FAQ

Questions are grouped into three buckets:

• General — applies regardless of runtime
• OpenClaw — questions specific to the OpenClaw runtime
• Hermes — questions specific to the Hermes runtime

For "what's the difference between OpenClaw and Hermes" and "which should I pick", see Runtime-OpenClaw and Runtime-Hermes.

---

General

Installation & system

Q: What are the system requirements?

macOS (Apple Silicon recommended) or Linux, with Docker. The install script handles Docker for you (Docker Desktop is recommended on Mac; otherwise Colima is auto-installed). Memory budget depends on what you run — see How much RAM do I need? below.

Q: Does it work on Intel Macs?

Yes, but Apple Silicon is recommended. ClawFleet uses your host architecture for Docker images.

Q: Do I need a cloud server?

No. ClawFleet runs entirely on your local machine. The only external cost is LLM API usage (or your existing ChatGPT subscription if you use Codex with OpenClaw).

Q: How much RAM do I need?

Idle memory, measured on M4 MacBook Air (16 GB RAM):

Instances	OpenClaw RAM	Hermes RAM
1	~700 MB	~140 MB
3	~2.1 GB	~400 MB
5	~3.5 GB	~700 MB

OpenClaw memory rises roughly 3× when the agent is actively browsing (Chromium loaded). Hermes stays roughly flat — it's headless. A mixed fleet of 3 OpenClaw + 3 Hermes is comfortable on 16 GB.

Configuration

Q: I can't save my model config — the button is greyed out.

You must click Test and see the "Validated" checkmark before saving. The save button is disabled until validation passes. (Codex skips this — OAuth completion counts as validation.)

Q: Can multiple instances share the same model config?

Yes. Models are shared across the entire fleet, regardless of runtime. One Anthropic config can power any number of OpenClaw + Hermes instances simultaneously.

Q: Can multiple instances share the same channel config?

No. Channels are exclusive — each channel config can only be assigned to one instance at a time. To run multiple instances on the same platform, register multiple bots and create one Channel Config per bot.

Q: What's the difference between Lark and the other channels?

Lark (Feishu) requires an App ID and App Secret instead of a single bot token. The Dashboard form adapts automatically when you select Lark. Lark is currently exposed for OpenClaw; for Hermes, configure it via Hermes's native Dashboard.

Q: Can I use a model not in the preset list?

Yes. When adding a model config, select "Custom..." from the model dropdown and type the model name. Make sure to test it before saving.

Operations

Q: How do I reset everything to a clean state?

make reset

This destroys all instances (both runtimes, including data), stops the Dashboard, and removes all build artifacts.

Q: Where is instance data stored?

Runtime	Host path	Inside container
OpenClaw	~/.clawfleet/data/<name>/openclaw/	/home/node/.openclaw
Hermes	~/.clawfleet/data/<name>/hermes/	/opt/data

Data survives container restarts and destroy (unless you use --purge).

Q: How do I reconfigure an instance?

Click Configure again on the instance card and select new model/channel configs. ClawFleet stops the existing gateway, applies the new configuration, and restarts it.

Q: My instance won't start — port conflict.

Each instance allocates two host ports based on its instance number: 690N and 1878N. If another process holds one of them, the instance can't start. Stop the conflicting process, or destroy and recreate the instance to get new port assignments.

Q: How do I view logs for debugging?

• CLI: clawfleet logs <name> -f to stream logs in real time (works for either runtime)
• Dashboard (OpenClaw): click Desktop on an OpenClaw instance card to see live logs on the detail page
• Dashboard (Hermes): open Hermes's native Dashboard via the ⚕ Dashboard button

Q: Can I run OpenClaw and Hermes side by side?

Yes — that's a primary use case. The Dashboard groups instances by runtime. They share the same Asset pool (model configs, channel configs) where the runtime supports it.

---

OpenClaw

Q: The image build is taking a long time.

The first OpenClaw image build downloads ~1.4 GB of layers. Subsequent builds are much faster due to Docker caching. You can also pull the pre-built image instead — that's what the install script does by default.

Q: My OpenClaw instance shows high memory usage.

OpenClaw includes a full XFCE desktop with Chromium. Memory usage rises significantly when Chromium is active (skills that drive a browser). If resources are tight, avoid opening the desktop view for multiple instances simultaneously.

Q: Can I access the OpenClaw desktop without the Dashboard?

Yes. Each OpenClaw instance exposes noVNC on port 690N (where N is the instance number). Open http://localhost:690N directly in your browser.

Q: My Codex login popup is blocked.

Some browsers block clawfleet.io's OAuth popup. Allow popups for clawfleet.io and re-click "Login with ChatGPT".

Q: Why does ClawFleet auto-set DM/group policies to "open"?

So you don't have to. When you assign a channel, ClawFleet runs OpenClaw's onboard with sensible defaults: DM policy "open", group policy "open", allowFrom: ["*"], channel plugin enabled, gateway started. You can override these later via the OpenClaw CLI inside the container, but the next time you click Configure the policies are reset.

---

Hermes

Q: How do I configure a Hermes instance?

Two layers:

1. ClawFleet's Configure dialog — quick model + channel setup (Discord, Telegram, or Slack). Gets the bot reachable in seconds.
2. Hermes's native Dashboard — click the ⚕ Dashboard button on the instance card. Opens Hermes's own UI for credential pools, cron, personality (SOUL.md), terminal backends, and channels beyond those three.

See Runtime-Hermes for the full walkthrough.

Q: The Hermes Dashboard button opens a port I don't recognise.

Hermes's native Dashboard listens on port 9119 inside the container. ClawFleet maps it to a host port allocated when the instance is created (typically 690N where N matches the instance number). The card always shows the current host port — read it from there, not from upstream documentation.

Q: Can I use my ChatGPT subscription with Hermes?

Not today. The Codex OAuth path is OpenClaw-exclusive in ClawFleet v1. For Hermes, use any standard API-key provider (Anthropic, Google AI Studio, OpenAI, DeepSeek) or configure long-tail providers (OpenRouter, Nous Portal, GLM, ...) via Hermes's native Dashboard.

Q: I want to chat with Hermes from my terminal.

clawfleet shell hermes-1

This drops you into Hermes's interactive TUI inside the container. Exit with /exit. The messaging Gateway keeps running independently — your bot stays online whether you're in the TUI or not.

Q: Why doesn't Hermes appear in the Soul Archive?

Soul Archive is OpenClaw-only today. Hermes parity is on the roadmap; until then, back up Hermes state by archiving ~/.clawfleet/data/<name>/hermes/ on the host.

Q: Why does ClawFleet's Configure for Hermes only list 3 channels?

The Dashboard scope today is Discord / Telegram / Slack. Other Hermes-supported platforms (WhatsApp, Signal, ...) work — they're configurable via Hermes's native Dashboard. Bringing them into ClawFleet's Configure dialog is tracked work toward Hermes parity.

Q: My Hermes instance lost its memory after I destroyed and recreated it.

Memory is per-instance. Destroying a Hermes instance removes its container; creating a new one starts from an empty memory layer. To preserve memory across instances today, archive ~/.clawfleet/data/<name>/hermes/ before destroying, and restore it before creating the replacement.