eva-security-audit（eva 安全审计）

OpenClaw Security Audit — BLUF Report

Non-interactive security snapshot for OpenClaw deployments. Runs openclaw security audit --deep, parses the output, and formats a structured BLUF report that can be sent to memory, Telegram, or stdout.

Security & Trust

This skill is a single SKILL.md file. You are reading its entire source right now.

• ✅ Read-only — does not write any files
• ✅ No network calls — only runs openclaw security audit --deep locally
• ✅ No credentials accessed — never reads API keys, tokens, or env vars
• ✅ No data exfiltration — nothing leaves your machine
• ✅ Passed OpenClaw security scan — verified clean on publish
• ✅ Transparent — this file is the complete skill. There is no code, binary, or script.

You can verify by running: openclaw security audit --deep yourself — this skill only formats that output.

Designed to be called by automation, agents, or cron — not a wizard. For interactive hardening (firewall, SSH, OS updates), use the healthcheck skill instead.

Workflow

1. Run the audit

openclaw security audit --deep

Capture full output. If running in background:

openclaw security audit --deep > /tmp/audit-$(date +%Y%m%d-%H%M).txt 2>&1

2. Parse findings

Extract every finding and classify by severity:

3. Assign posture rating

4. Format the BLUF report

Produce this exact structure — fill in real values, omit empty sections:

SECURITY AUDIT — YYYY-MM-DD HH:MM
POSTURE: [GREEN/YELLOW/RED] — X critical · Y warn · Z info

BLUF: [One sentence: overall risk and the single most important action.]

CRITICAL
1. [finding-id] [Description — blast radius]
   Fix: [exact command or config change]
2. ...

WARN
1. [finding-id] [Description]
   Fix: [exact command or config change]

INFO
- [finding-id] [Context note]

NEXT STEPS
1. Apply fixes above (copy-paste ready).
2. Re-run: openclaw security audit --deep
3. Log findings: append to memory/YYYY-MM-DD.md

Audit complete. Re-run after each fix to confirm POSTURE GREEN.

Rules for the report:

• BLUF first, always. One sentence max.
• Every CRITICAL must have a copy-paste fix command.
• Never omit a CRITICAL finding, even if it seems minor.
• If 0 findings: state "POSTURE: GREEN — 0 critical · 0 warn" and stop.
• Do not add commentary, context, or suggestions beyond what the audit output contains.

5. Deliver the report

Choose one or more delivery targets based on user context:

Memory (default for scheduled runs):

# Append to today's memory file
echo "[audit result]" >> memory/$(date +%Y-%m-%d).md

Telegram (if BOT_TOKEN and CHAT_ID are in environment):

import os, requests
requests.post(
    f"https://api.telegram.org/bot{os.getenv('TELEGRAM_BOT_TOKEN')}/sendMessage",
    json={"chat_id": os.getenv('MASTER_TELEGRAM_ID'), "text": report}
)

Stdout only: print the report and exit.

Quick-fix reference

These are the most common findings and their fixes. Apply and re-run to confirm:

Apply all CRITICAL fixes first, then re-run before addressing WARNs.

Scheduling (cron)

To run this audit automatically (e.g. daily at 04:00):

openclaw cron add --name "security-audit:daily" --cron "0 4 * * *" --prompt "Run the eva-security-audit skill and send the report to memory and Telegram."

Check scheduled jobs:

openclaw cron list

Notes

• This skill is read-only. It does not modify any config, firewall, or SSH settings.
• For --fix (applies OpenClaw safe defaults automatically): openclaw security audit --deep --fix

Confirm impact before running --fix in production.

• For JSON output suitable for piping: openclaw security audit --deep --json
• For interactive OS hardening (firewall, SSH, updates): use the healthcheck skill.