Tools

Clawvitals

Name: Clawvitals
Rating: 3.5 (1 reviews)
Author: ANGUARDA

By ANGUARDA 👁 36 views ▲ 0 votes

Security vitals checker for OpenClaw, providing enhanced checks and functionality beyond the ClawVitals skill (https://clawhub.ai/bk-cm/clawvitals). Checks your installation for common misconfigurations and security issues, with scheduled scans, delta detection, regression alerts, and scan history. Full details: https://clawvitals.io/plugin

GitHub

Install

openclaw plugins install clawhub:claw-security-vitals

Configuration Example

{
  "controls": { "mode": "expanded" }
}

README

# ClawVitals Plugin

Security vitals checker for self-hosted [OpenClaw](https://openclaw.ai) installations. Recurring security checks, scan history, delta detection, and regression-aware alerting.

> This is the **plugin** — the stateful, scheduled, telemetry-enabled upgrade from the [ClawVitals Skill](https://clawhub.ai/bk-cm/clawvitals).

---

## Contents

- [Skill vs Plugin](#skill-vs-plugin)
- [Install](#install)
- [Uninstall](#uninstall)
- [Commands](#commands)
- [Agent tools](#agent-tools)
- [Standard vs Expanded controls](#standard-vs-expanded-controls)
- [Example output](#example-output)
- [Regression alerts](#regression-alerts)
- [Scheduling](#scheduling)
- [Fleet Management](#fleet-management)
- [Exclusion management](#exclusion-management)
- [Telemetry](#telemetry)
- [Configuration](#configuration)
- [Directory structure](#directory-structure)
- [License](#license)

---

## Skill vs Plugin

The **ClawVitals skill** (on ClawHub) is stateless — it runs a point-in-time scan, prints the result, and stores nothing. No telemetry, no network calls, no persistent state. It is locked and will not change.

The **plugin** is the upgrade path. It adds everything the skill deliberately omits:

| Feature | Skill | Plugin |
|---|---|---|
| Scan & score | ✅ | ✅ |
| Remediation steps | ✅ | ✅ |
| Experimental controls | ✅ | ✅ |
| Scan history & delta detection | ❌ | ✅ |
| Recurring scheduled scans | ❌ | ✅ |
| Regression + critical alerts | ❌ | ✅ |
| Exclusion management | ❌ | ✅ |
| Scan history on dashboard (coming soon) | ❌ | ✅ |
| Fleet management (alias) | ❌ | ✅ |
| Telemetry | none | **on by default (opt-out)** |

---

## Install

ClawVitals Plugin is published on [ClawHub](https://clawhub.ai/plugins/claw-security-vitals).

```bash
openclaw plugins install clawhub:claw-security-vitals
```

After installing, run your first scan:

```
run clawvitals
```

---

## Uninstall

```bash
openclaw plugins uninstall claw-security-vitals
```

After uninstalling, `run clawvitals` will fall back to the skill if it is still installed, or return a "not found" error if neither is installed.

> **Note:** Uninstalling does not delete your scan history. Run files are stored at `{workspace}/clawvitals/runs/` and are retained according to your configured retention policy (default: 90 days). To remove all data, delete this directory manually.

---

## Commands

These are chat commands you type directly in your OpenClaw messaging surface:

| Command | Description |
|---|---|
| `run clawvitals` | Run a full security scan (standard controls) |
| `run clawvitals --expanded` | Run scan with expanded system-level controls (see [expanded controls](#standard-vs-expanded-controls)) |
| `run clawvitals --standard` | Run scan with standard controls only (explicit) |
| `show clawvitals details` | Full report with all findings and remediation steps |
| `clawvitals status` | Show last scan time, score, schedule, and trial/plan status |
| `clawvitals help` | Show command reference |

---

## Agent tools

The following tools are invoked by the agent (not typed as chat commands). You can trigger them via natural language — for example, say "set clawvitals schedule to daily" and the agent will call the appropriate tool.

| Tool | Description |
|---|---|
| `clawvitals_set_alias` | Set a friendly name for this host in reports and dashboard |
| `clawvitals_show_identity` | Show install UUID, alias, and dashboard link |
| `clawvitals_telemetry` | Enable or disable telemetry |
| `clawvitals_set_schedule` | Configure recurring scan cadence |
| `clawvitals_status` | Show current status |
| `clawvitals_trial_status` | Show trial status and upgrade options |
| `clawvitals_upgrade` | Upgrade to a paid plan |
| `clawvitals_configure_webhook` | Set up a webhook for alert delivery |
| `clawvitals_exclude` | Suppress a finding with a reason |
| `clawvitals_list_exclusions` | List all active exclusions |
| `clawvitals_remove_exclusion` | Remove an exclusion |
| `clawvitals_get_report` | Retrieve a scan report |
| `clawvitals_approve_cognitive_file` | Approve a cognitive file |

---

## Standard vs Expanded controls

By default the plugin runs in **standard mode** — the same OpenClaw-native control set as the skill, plus scan history, delta detection, and alerting. Standard mode uses only the OpenClaw CLI (`openclaw security audit`, `openclaw health`, etc.) and requires no additional permissions.

**Expanded mode** adds a second layer of system-level checks that require direct filesystem and shell access. These are the checks the skill can never do.

### Switch to expanded mode

```
run clawvitals --expanded         # one-off expanded scan
run clawvitals --standard         # one-off standard scan (explicit default)
```

Or set it as your default via `openclaw.plugin.json`:
```json
{
  "controls": { "mode": "expanded" }
}
```

### What expanded mode adds

| ID | Control | Severity | What it checks |
|---|---|---|---|
| **NC-OLLAMA-001** | Ollama not externally accessible | 🔴 Critical | Checks whether Ollama is running and if port 11434 is bound to a public interface. 175,000+ exposed Ollama instances found in 2026 — active "LLMjacking" attacks target this. |
| **NC-NET-001** | Management interfaces not internet-exposed | 🔴 Critical | Scans open ports for SSH (22), Docker API (2375/2376), and common admin dashboards (8080, 9000) and checks whether they're reachable beyond localhost. |
| **NC-SECRET-001** | No secrets in env/config files | 🔴 Critical | Regex-scans `~/.env`, `.envrc`, and common config files for API key patterns. The most common cause of credential compromise. |
| **NC-SECRET-002** | No API keys in shell history | 🟠 High | Scans `~/.zsh_history` and `~/.bash_history` for secret patterns (API keys, tokens, passwords passed as arguments). Commonly overlooked. |
| **NC-TUNNEL-001** | Cloudflare tunnel endpoints authenticated | 🟠 High | Checks `~/.cloudflared/` config to confirm tunnel-exposed services require authentication. Unauthenticated tunnels are an open door. |
| **NC-DOCKER-001** | Containers not running as root or privileged | 🟠 High | Runs `docker inspect` on running containers to check for `--privileged`, root user, or dangerous capability grants. Aligns with CIS Docker Benchmark. |
| **NC-OS-001** | OS auto-updates enabled | 🟠 High | Checks that automatic OS updates are enabled (`softwareupdate` on macOS, `unattended-upgrades` on Linux). Often neglected on self-hosted machines. |
| **NC-OS-002** | Disk encryption enabled | 🟠 High | Checks FileVault status (macOS) or LUKS encryption (Linux). Critical for Mac Mini and home server deployments where physical access is a real risk. |

All expanded checks are **read-only** — nothing is modified. See [SECURITY.md](./SECURITY.md) for the full list of commands and file paths accessed.

### Expanded mode output

When expanded mode runs, the report clearly labels the section:

```
ClawVitals Plugin v1.0.1 🔌  ·  Expanded Scan

━━━ STANDARD CONTROLS ━━━━━━━━━━━━━━━━━━━━━
[standard control results — see example output below]

━━━ EXPANDED CONTROLS ━━━━━━━━━━━━━━━━━━━━━

🔴 CRITICAL  NC-OLLAMA-001  Ollama externally accessible
Evidence: Port 11434 bound to 0.0.0.0 — accessible from outside localhost
Fix: Set OLLAMA_HOST=127.0.0.1 in your Ollama environment and restart:
     launchctl setenv OLLAMA_HOST "127.0.0.1"   # macOS
     systemctl edit ollama                        # Linux (add Environment=OLLAMA_HOST=127.0.0.1)
→ https://clawvitals.io/docs/NC-OLLAMA-001

🟠 HIGH  NC-SECRET-002  API key pattern found in shell history
Evidence: Pattern matching sk-... found in ~/.zsh_history (line ~342)
Fix: Run `history -c` to clear in-memory history, then manually edit ~/.zsh_history
     to remove the line. Rotate the exposed key immediately.
→ https://clawvitals.io/docs/NC-SECRET-002

✅ NC-NET-001    No management interfaces exposed
✅ NC-SECRET-001  No secrets found in env/config files
✅ NC-TUNNEL-001  Cloudflare tunnel endpoints authenticated
✅ NC-DOCKER-001  Containers not privileged
✅ NC-OS-001     Auto-updates enabled
✅ NC-OS-002     Disk encryption enabled (FileVault ON)

Expanded score: 2 new findings  ·  6 passed
```

---

## Example output

### Summary message (after `run clawvitals`)

```
ClawVitals Plugin v1.0.1 🔌

🔴 Security Score: 58 / 100  ·  RED
Host: mac-mini-home  ·  Scanned: 2026-04-15 15:38 BST

Findings: 2 Critical  ·  1 High  ·  1 Medium
Delta: ▲ 1 new finding since last scan (2026-04-08)

─────────────────────────────────────────
CRITICAL  NC-OC-012  Gateway auth disabled
CRITICAL  NC-OC-003  Command policy: deny-only mode
HIGH      NC-VERS-001  OpenClaw update available (2026.3.13 → 2026.4.1)
MEDIUM    NC-OC-008  Channel health degraded

▶ Reply "show clawvitals details" for full report with remediation steps.
📈 Track your scans → https://clawvitals.io/dashboard
```

### Full details (after `show clawvitals details`)

```
ClawVitals Plugin v1.0.1 🔌  ·  Full Report
Host: mac-mini-home  ·  Control Library v1.0.1  ·  OpenClaw 2026.3.13

━━━ CRITICAL ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

[NC-OC-012] Gateway auth disabled
Severity: Critical  ·  Source: security_audit (authoritative)
Evidence: groups[0].auth.type = "none"
Fix: openclaw gateway auth set --type bearer --token <your-token>
Docs: https://clawvitals.io/docs/NC-OC-012

[NC-OC-003] Command policy: deny-only mode
Severity: Critical  ·  Source: security_audit (authoritative)
Evidence: commandPolicy = "deny"
Fix: openclaw policy set --commands allowlist
Docs: https://clawvitals.io/docs/NC-OC-003

━━━ HIGH ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

[NC-VERS-001] OpenClaw update available
Severity: High  ·  Source: update_status (authoritative)
Evidence: current=2026.3.13, latest=2026.4.1, channel=stable
Fix: openclaw update
Docs: https://clawvitals.io/docs/NC-VERS-001

━━━ MEDIUM ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

[NC-OC-008] Channel health degraded
Severity: Medium  ·  Source: health (contextual)
Evidence: channels[0].status = "degraded"
Fix: Ch

... (truncated)

tools