result = moderate("加V：xxxx，免费领取项目源码，速来！")

print(result)

期望输出类似：{'verdict': 'review', 'reason': '含广告推广内容'}

Once this code runs, you've completed the core pipeline: comment text → whitelist judgment → structured result.

### Step 4: Integrate with Your Blog

The recommended architecture is simple:

1. After a comment is submitted, it first goes through a local blacklist (regex quick filter — zero cost, intercepts the most obvious spam);
2. Comments that pass the quick filter call the `moderate()` function above; those marked `approve` are published directly;
3. Comments marked `review` go into a pending queue, which you can process in one batch daily.

If your blog is a static site with no backend, you can wrap this in a cloud function (such as Vercel Functions). Be careful not to expose the API Key to the frontend.

### Step 5: Tips for Controlling Cost and Latency

- **Model tiering**: Use a lightweight model for regular comments, and only escalate long texts or uncertain cases to a stronger model — this significantly reduces per-comment moderation cost;
- **Add timeouts and fallbacks**: When the AI service occasionally times out, fall back to "send to the manual queue" rather than letting the comment API hang;
- **Cache duplicate content**: Reuse the previous result for repeated comments from the same IP within a short time window.

## Before-and-After Comparison

Here's my ledger before and after the integration:

| Item | Before | After |
|---|---|---|
| Daily moderation time | ~40 minutes | Under 5 minutes |
| Weekly rule maintenance | 1–2 hours | Essentially zero |
| New spam-type interception | Days of lag | Effective same-day |

All told, this saves about **20 hours** of manual work per month, while the API cost — per the official pricing page — is far less than the value of that time for me. For an indie developer, this might be the highest-ROI automation upgrade you can make.

## Final Thoughts

The barrier to entry for whitelist-based moderation is far lower than you'd imagine: register, grab a key, twenty lines of code, and you can go live in one evening. If you also want to turn your comment section from "a daily burden" into "an automated pipeline," go register an account first and get the first piece of code in this article running: https://api.thistoken.ai/register

---

Ready to try it yourself? Sign up at https://api.thistoken.ai/register to get your API key and start building.