Google AI Usage Monitor

Monitor Google AI Studio (Gemini API) usage, rate limits, and quota consumption with automated alerts.

Security Scan

The skill's instructions match the stated monitoring goal but omit how it will authenticate to Google and Discord and rely on using a browser profile (potentially exposing full Google session), which is an incoherence that increases risk.

Purpose & Capability

The skill claims to monitor Google AI Studio usage via browser automation and send alerts to Discord — that purpose is plausible. However, the skill requests no credentials or webhooks yet instructs accessing a protected Google usage dashboard and posting to a Discord channel. Legitimate implementations would normally require explicit authentication details (OAuth/service account, API credentials, or a Discord webhook/token). The omission is inconsistent with the described capability.

Instruction Scope

SKILL.md directs the agent to open the Google AI Studio usage page using a named browser profile (profile=openclaw) and parse live dashboard data. That implies the agent will access the user's authenticated Google session and potentially any data available in that session. The instructions do not limit access to only usage metrics and do not describe how to obtain consent or restrict scope. They also assume the ability to post to Discord without specifying how authentication is performed.

Install Mechanism

This is an instruction-only skill with no install steps and no code files, so there is no download/execute footprint to review. That lowers supply-chain risk, but runtime actions (browser automation, posting to external channels) remain significant.

Credentials

No required environment variables or primary credential are declared, yet the workflow clearly needs authenticated access to a Google account/project and a way to send Discord messages (webhook or bot token). The absence of declared credentials is disproportionate and unclear — it suggests the skill expects to reuse existing browser sessions or platform-level secrets without explicitly stating or limiting them.

Persistence & Privilege

always:false and standard autonomous invocation are fine. The SKILL.md includes a cron job example and delivery instructions for posting to Discord, which implies autonomous periodic checks. Autonomous invocation combined with implicit use of an agent browser profile increases the blast radius if credentials/sessions are broad, so you should be cautious even though no 'always:true' or installer-level persistence is requested.

What to consider before installing

Before installing or enabling this skill, ask the author how authentication is performed and avoid giving it access to your primary Google session. Prefer a least-privilege approach: use an explicit service account or monitoring API (not dashboard scraping) with only read/monitoring permissions, and provide a scoped Discord webhook or bot token dedicated to alerts. Do not rely on a shared browser profile (profile=openclaw) unless you understand exactly which account/session it will use; that profile could expose unrelated Google data. If you must use this skill, run it in an isolated agent/session, insist the skill declare required env vars (Google credentials or service account, DISCORD_WEBHOOK_URL or bot token), and review any actual code (if provided) to confirm it only reads the intended dashboard endpoints and only posts to the specified channel. If the source/author is unknown or you cannot supply scoped credentials, treat this skill as risky and avoid installing it.

Google AI Usage Monitor Skill

Monitor Google AI Studio usage to prevent quota exhaustion and optimize API consumption.

Supported Metrics

Metric	Description	Alert Threshold
RPM	Requests Per Minute (peak)	> 80% of limit
TPM	Tokens Per Minute (peak)	> 80% of limit
RPD	Requests Per Day	> 80% of limit

Rate Limits by Tier

Tier	Typical Limits
Free	2 RPM, 32K TPM, 50 RPD
Pay-as-you-go	10-15 RPM, 100K+ TPM, 500+ RPD
Paid Tier 1	20 RPM, 100K TPM, 250 RPD (varies by model)

Note: Actual limits vary by model and can be viewed at the usage dashboard.

Usage Dashboard

URL

https://aistudio.google.com/usage?project={PROJECT_ID}&timeRange=last-28-days&tab=rate-limit

Key Elements to Extract

• Project name: Which GCP project
• Tier: Free / Pay-as-you-go / Paid tier X
• Models table: Each row contains model name, category, RPM, TPM, RPD
• Time range: Default 28 days

Browser Automation

Open Usage Page

// Using OpenClaw browser tool
browser action=open targetUrl="https://aistudio.google.com/usage?project=YOUR_PROJECT_ID&timeRange=last-28-days&tab=rate-limit" profile=openclaw

Wait for Data Load

The page loads data asynchronously. Wait for:

1. Project dropdown shows project name (not "Loading...")
2. Rate limits table has data rows

Parse Table Data

Look for table rows with pattern:

Model Name | Category | X / Y | X / Y | X / Y | View in charts

Where X / Y represents used / limit.

Report Format

Discord Message Template

## 📊 Google AI Studio 用量报告

**项目**: {project_name}
**付费等级**: {tier}
**统计周期**: 过去 28 天

---

### {Model Name}
| 指标 | 用量 | 限额 | 使用率 |
|------|------|------|--------|
| RPM | {rpm_used} | {rpm_limit} | {rpm_pct}% |
| TPM | {tpm_used} | {tpm_limit} | {tpm_pct}% |
| RPD | {rpd_used} | {rpd_limit} | {rpd_pct}% |

---

{status_emoji} **状态**: {status_text}

*检查时间: {timestamp}*

Status Levels

Usage %	Status	Emoji	Action
< 50%	正常	✅	Continue normally
50-80%	需关注	⚠️	Monitor more frequently
> 80%	风险预警	🚨	Alert user, consider rate limiting

Alert Rules

When to Alert User

1. Any metric > 80%: Immediate alert with @mention
2. Any metric > 50%: Include warning note in report
3. API errors (429): Track rate limit hits

Alert Message Template

🚨 **Google AI 配额预警**

<@USER_ID> 以下指标接近限额：

- **{model}** {metric}: {used}/{limit} ({pct}%)

建议：
- 减少 API 调用频率
- 考虑升级付费等级
- 检查是否有异常调用

Cron Job Setup

Daily Check (Recommended)

{
  "name": "Google AI 用量检查",
  "schedule": {
    "kind": "cron",
    "expr": "0 20 * * *",
    "tz": "Asia/Shanghai"
  },
  "sessionTarget": "isolated",
  "payload": {
    "kind": "agentTurn",
    "message": "检查 Google AI Studio 用量并发送报告到指定 Discord 频道"
  },
  "delivery": {
    "mode": "announce",
    "channel": "discord",
    "to": "CHANNEL_ID"
  }
}

Integration with OpenClaw

Configuration

Add to TOOLS.md:

## Google AI Studio

- **Project ID**: gen-lang-client-XXXXXXXXXX
- **Dashboard**: https://aistudio.google.com/usage
- **Discord Channel**: #google-ai (CHANNEL_ID)
- **Check Schedule**: Daily 20:00

Heartbeat Integration

Add to HEARTBEAT.md:

## Google AI Monitoring
- Check usage if last check > 24 hours
- Alert if any metric > 80%

Troubleshooting

Page Not Loading

1. Check if logged into correct Google account
2. Verify project ID is correct
3. Wait longer for async data load (5-10 seconds)

Data Shows "Loading..."

The project dropdown may take time to populate. Retry snapshot after a few seconds.

Metrics Not Updating

Google notes: "Usage data may take up to 15 minutes to update."

References

• Google AI Studio Usage Dashboard
• Gemini API Rate Limits
• Billing Documentation
• Cloud Monitoring for Gemini