revsec-shield（转速保护罩）

RevSec Shield

One-line description: 24/7 security monitoring for your OpenClaw agent — delivered to your WhatsApp in plain English.

Overview

• What it does: Monitors your OpenClaw agent for prompt injection attacks,

malicious skills, and unexpected configuration changes. Sends WhatsApp alerts when something suspicious is detected.

• When to use it: Install once. It runs silently in the background. You only

hear from it when something needs your attention.

• Requirements: A free RevSec account (revsec.revt2d.com/signup), WhatsApp

connected to OpenClaw, and a REVSEC_API_KEY environment variable.

RevSec Shield is a monitoring and alerting tool. It detects threats and puts them in front of you immediately. You stay in control of what happens next.

Three operating modes:

1. Setup — first-time configuration, run once, never needs repeating
2. Alert poll — background cron job, runs every 5 minutes silently
3. Manual check — user asks for a status update

Quick Start

Install

Ask your OpenClaw agent:

Use clawhub to install revsec-shield

Or via CLI:

npm i -g clawhub
clawhub install revsec-shield

Configure

Add your API key in OpenClaw's environment settings (not shell exports):

REVSEC_API_KEY=rsk_personal_<your_key>

Get your free key at: https://revsec.revt2d.com/signup

Activate

Ask your OpenClaw agent:

Run the revsec-shield setup

That's it. RevSec Shield will confirm activation and start monitoring immediately.

Environment Variable Contract

Set via OpenClaw's environment settings UI or .env file — never as a shell export.

Core rules (always follow)

• Never ask the user to do something manually that can be done via curl or shell.
• RevSec detects first, reports second. Never ask the user "should I flag this?" —

report what was already detected.

• Keep alert messages plain and human. No JSON, no confidence scores, no

technical jargon unless the user asks for detail.

• State file is ~/.openclaw/revsec-state.json. Read before every action.

Write after every action that changes state.

• If the state file is missing, run Setup before anything else.
• If REVSEC_API_KEY is not set, prompt the user to get one and stop.
• All curl calls must include -s (silent) and -f (fail on HTTP errors).
• On any curl failure: log the error to state, stay silent to the user unless

it is a setup step or they explicitly asked for a status check.

• Never re-run setup if it has already completed. Use the existing

openclaw_agent_id from state for all subsequent API calls.

State file — ~/.openclaw/revsec-state.json

Read and write this file to persist identity and poll state across sessions.

Schema:

{
  "openclaw_agent_id": "<uuid — generated once, never changes>",
  "agent_id": "<uuid — returned by RevSec on registration>",
  "org_id": "<uuid — returned by RevSec on registration>",
  "last_poll_at": "<ISO 8601 timestamp — advanced after each successful poll>",
  "registered_at": "<ISO 8601 timestamp>",
  "registered_skills": ["skill1", "skill2"],
  "skill_version": "1.0.0",
  "setup_complete": true
}

Read state:

cat ~/.openclaw/revsec-state.json 2>/dev/null || echo "{}"

Write state (always write the full object, never partial):

cat > ~/.openclaw/revsec-state.json << 'EOF'
{ <full updated state object> }
EOF

Workflow 1 — Setup (run when setup_complete is missing or false)

BEFORE ANYTHING ELSE — Check if already set up:

Read state file:

cat ~/.openclaw/revsec-state.json 2>/dev/null || echo "{}"

If setup_complete is true:

1. Check if cron job exists: openclaw cron list
2. If revsec:alert-poll is missing → create it (Step 5 only)
3. Tell user: "RevSec Shield is already active and protecting your agent."
4. STOP — do not proceed with Steps 1–4. Do not re-register.

Only run Steps 1–6 if setup_complete is missing or false.

Step 1 — Check for API key

echo $REVSEC_API_KEY

If empty or unset, stop and tell the user:

To activate RevSec Shield, get your free API key at https://revsec.revt2d.com/signup Then add it to your OpenClaw environment settings: REVSEC_API_KEY=rsk_personal_<your_key>

Do not proceed until the key is set.

Step 2 — Generate stable agent identity

Generate a UUID. This is the openclaw_agent_id — it never changes.

python3 -c "import uuid; print(uuid.uuid4())"

If python3 is unavailable:

cat /proc/sys/kernel/random/uuid 2>/dev/null || uuidgen

Store this value. It will be written to state in Step 4.

Step 3 — Collect agent config from OpenClaw environment

Gather what is available from the local environment. Do not ask the user questions that can be inferred.

# Hostname
hostname

# OpenClaw model currently configured (read from config)
cat ~/.openclaw/openclaw.json 2>/dev/null | python3 -c "
import json,sys
d=json.load(sys.stdin)
print(d.get('model','unknown'))
" 2>/dev/null || echo "unknown"

# Skills currently installed (list skill directories)
ls ~/.openclaw/skills/ 2>/dev/null || ls ~/clawd/skills/ 2>/dev/null || echo ""

Capture the skills list — this becomes the baseline for change detection.

Step 4 — Register agent with RevSec

Construct the registration payload from the collected values and call the API.

curl -sf -X POST \
  "https://revsec.revt2d.com/fcc/api/v1/personal/register-agent" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $REVSEC_API_KEY" \
  -d '{
    "openclaw_agent_id": "<uuid from Step 2>",
    "skills": [<comma-separated list of installed skill names as strings>],
    "model": "<model from Step 3>",
    "channels": ["openclaw"],
    "integrations": [<any external services visible in openclaw.json>],
    "hostname": "<hostname from Step 3>",
    "skill_version": "1.0.0"
  }'

On success the API returns { "agent_id": "...", "action": "registered" }.

Write the full state file including the skills baseline:

cat > ~/.openclaw/revsec-state.json << 'EOF'
{
  "openclaw_agent_id": "<uuid from Step 2>",
  "agent_id": "<agent_id from API response>",
  "org_id": "<org_id from API response if present, else empty string>",
  "last_poll_at": "<current UTC timestamp in ISO 8601>",
  "registered_at": "<current UTC timestamp in ISO 8601>",
  "registered_skills": [<same skills list sent to API>],
  "skill_version": "1.0.0",
  "setup_complete": true
}
EOF

Step 5 — Schedule the background alert poll

Check if the cron job already exists:

openclaw cron list

Look for a job named revsec:alert-poll. If it exists, skip creation.

If it does not exist, create it:

openclaw cron add \
  --name "revsec:alert-poll" \
  --model "google/gemini-2.0-flash" \
  --schedule "*/5 * * * *" \
  --prompt "Run the revsec-shield skill in alert poll mode. Do these two things silently: 1) Check if any new skills have been installed since last registration by comparing the current skills list against the registered_skills array in the state file — if changed, call register-agent automatically with the updated skills list and update registered_skills in state. 2) Poll for new security alerts from RevSec since the last poll timestamp and deliver any alerts to the user. Be completely silent if there are no changes and no alerts."

Step 6 — Confirm setup to the user

Tell the user exactly this (keep it short, match the tone):

✅ RevSec Shield is active. Your agent is registered and protected. I'll check for threats every 5 minutes and alert you here if anything is detected. You won't hear from me unless something happens. Your dashboard: https://revsec.revt2d.com/personal 📱 If you use WhatsApp: Send me a message from WhatsApp now to activate alert delivery. Without this first message, I cannot reach you on WhatsApp when threats are detected. Just say "Hi" from the number you connected during setup.

Workflow 2 — Alert poll (triggered by cron job revsec:alert-poll)

This workflow runs every 5 minutes automatically. It must be fast and silent when there are no changes and no alerts. Never output anything to the user unless there is something they need to know.

Step 1 — Read state

cat ~/.openclaw/revsec-state.json

If state is missing or setup_complete is not true, output nothing and stop. Do not trigger setup from a cron context.

Step 1b — Check for config changes (automatic, silent)

Get the current installed skills:

ls ~/.openclaw/skills/ 2>/dev/null || ls ~/clawd/skills/ 2>/dev/null || echo ""

If registered_skills is missing from state file: Treat it as an empty list. All current skills are "new". Re-register once to set the baseline, then write registered_skills to state. Stay silent — this is a one-time migration for users upgrading from an older version.

If the skills list has changed (new skill added or skill removed):

1. Re-register the agent using the existing openclaw_agent_id from state:

curl -sf -X POST \
  "https://revsec.revt2d.com/fcc/api/v1/personal/register-agent" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $REVSEC_API_KEY" \
  -d '{
    "openclaw_agent_id": "<openclaw_agent_id from state — never changes>",
    "skills": [<current skills list>],
    "model": "<model from openclaw.json or unknown>",
    "channels": ["openclaw"],
    "integrations": [],
    "hostname": "<hostname>",
    "skill_version": "1.0.0"
  }'

1. Update registered_skills in the state file with the new skills list.

Keep all other state values unchanged.

1. Stay completely silent — this is automatic background maintenance.

The RevSec policy engine will evaluate the updated config and create violations if any new skills are suspicious. Those violations will be picked up in Step 2 on this same poll cycle.

If the skills list has not changed: skip to Step 2.

Step 2 — Poll for new alerts

curl -sf \
  "https://revsec.revt2d.com/fcc/api/v1/personal/alerts?since=<last_poll_at from state>" \
  -H "Authorization: Bearer $REVSEC_API_KEY"

If curl fails (network error, timeout, 5xx): update nothing, stay silent, stop. Do not advance last_poll_at on failure — the same window will be retried on the next poll.

If the response is an empty array []: stay completely silent. Advance last_poll_at to now and write state. Stop.

Step 3 — Deliver alerts (only if response is non-empty)

For each alert in the response, use the message field from the API response directly — it is already formatted for human delivery. Do not add your own headers or reformat it. Do not prepend "RevSec blocked a threat" — the message field already contains the full alert text.

If there are multiple alerts, deliver them as a single message grouped by severity — critical first, then high, then medium. Use a separator line between groups only if there are multiple severity levels.

Example output format:

<message field from API — critical alerts first>

──────────────────────────

<message field from API — medium/low alerts>

Step 4 — Advance poll timestamp

After successful delivery (or confirmed empty response), update last_poll_at to the current UTC time and write the full state file.

python3 -c "from datetime import datetime,timezone; print(datetime.now(timezone.utc).isoformat())"

Workflow 3 — Manual security check (user asks "how am I protected?" or "what did RevSec detect?")

Step 1 — Read state

Check ~/.openclaw/revsec-state.json. If setup is not complete, run Workflow 1 first.

Step 1b — Check cron health (self-healing)

openclaw cron list

If revsec:alert-poll is missing from the list — recreate it silently using Step 5 from Workflow 1, then tell the user:

Background monitoring was inactive — I've restarted it. Your protection is now fully active again.

Step 2 — Fetch dashboard summary

curl -sf \
  "https://revsec.revt2d.com/fcc/api/v1/personal/summary" \
  -H "Authorization: Bearer $REVSEC_API_KEY"

Step 3 — Report to user

Use the summary data to report in plain language. Follow this format exactly:

🛡️ RevSec Shield — last 7 days

Threats detected: <threats_blocked_7d>
Flagged for review: <threats_flagged_7d>
Agent: <agent.name> (<agent.agent_framework>)
Status: <agent.status>
Policies active: <active_policies>

Full details: https://revsec.revt2d.com/personal

If threats_blocked_7d and threats_flagged_7d are both 0:

🛡️ RevSec Shield — no threats detected in the last 7 days.
Your agent is clean. Protection is active.

Workflow 4 — Config update (automatic via cron, or triggered manually)

This workflow is called automatically by the cron job when a config change is detected. It can also be triggered manually if the user explicitly asks to update their RevSec registration.

The user never needs to run this manually — it happens automatically.

Step 1 — Read current state

Get openclaw_agent_id from ~/.openclaw/revsec-state.json.

Step 2 — Collect current config

# Current skills
ls ~/.openclaw/skills/ 2>/dev/null || echo ""

# Current model
cat ~/.openclaw/openclaw.json 2>/dev/null | python3 -c "
import json,sys
d=json.load(sys.stdin)
print(d.get('model','unknown'))
" 2>/dev/null || echo "unknown"

# Hostname
hostname

Step 3 — Re-register with updated config

curl -sf -X POST \
  "https://revsec.revt2d.com/fcc/api/v1/personal/register-agent" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $REVSEC_API_KEY" \
  -d '{
    "openclaw_agent_id": "<openclaw_agent_id from state — never changes>",
    "skills": [<current skills list>],
    "model": "<current model>",
    "channels": ["openclaw"],
    "integrations": [],
    "hostname": "<hostname>",
    "skill_version": "1.0.0"
  }'

Step 4 — Update state file

Update registered_skills in the state file with the new skills list. Keep all other state values unchanged — especially openclaw_agent_id.

If triggered manually by the user, confirm:

✅ RevSec updated — your agent profile is now in sync.

If triggered automatically by cron, stay silent.

Security & Guardrails

• No message content sent: RevSec only receives agent metadata — skill names,

hostname, model. Your conversations and message content are never sent to RevSec.

• Credentials: API key stored in OpenClaw runtime environment only.

Never logged or transmitted beyond the RevSec API.

• Network access: Only connects to revsec.revt2d.com. No other outbound

connections are made by this skill.

• Local storage: State stored only in ~/.openclaw/revsec-state.json.

No other local files are created or modified.

• Fail-closed on API errors: If RevSec API is unreachable, skill stays

silent and retries on next poll. Agent behaviour is never modified.

• No pipe-to-interpreter: No curl | bash, curl | python, or similar

patterns are used anywhere in this skill.

• No credential harvesting: This skill does not read environment variables

beyond REVSEC_API_KEY, does not access credential files, and does not transmit system information beyond hostname and skill names.

• Read-only agent monitoring: This skill never modifies OpenClaw settings,

agent configuration, or installed skills.

Error handling

Troubleshooting

What RevSec Shield does NOT do

• It does not change any OpenClaw settings or configuration
• It does not access any data beyond what the agent config exposes
• It does not store any data locally beyond ~/.openclaw/revsec-state.json
• It does not send prompts or conversation content to RevSec — only metadata
• It does not block or modify LLM calls directly — it monitors agent

configuration and behaviour via the RevSec API

• It does not initiate WhatsApp sessions — user must send first message

Cron job reference

Use openclaw cron list to verify the job is registered. Use openclaw cron runs revsec:alert-poll to see recent run history.

Release notes

v1.0.0 — Initial public release

• Agent registration and monitoring
• 5-minute background polling
• WhatsApp alert delivery
• Automatic config change detection
• New integration alerts
• Personal security dashboard at revsec.revt2d.com/personal

Links

• Dashboard: https://revsec.revt2d.com/personal
• Free signup: https://revsec.revt2d.com/signup
• Team plan: https://revsec.revt2d.com/signup/team
• Support: hello@revupai.com

Publisher

• Publisher: @RevUp-AI
• Homepage: https://revsec.revt2d.com
• Support: hello@revupai.com
• GitHub: https://github.com/RevUp-AI/revsec-shield