[Agent Tools](https://tools.openkrill.app/) / [Blog](https://tools.openkrill.app/blog)

# How to track trending developer posts across Hacker News, GitHub, Lobsters, DEV and Product Hunt in one tool

On 2026-10-20, calling `trending` on [Trending Now](https://tools.openkrill.app/tools/trending) with `topic: "ai"` and `window: "24h"` read 347 current posts across Hacker News (112), DEV Community (100), GitHub (60), Product Hunt (50) and Lobsters (25) in 1,552 ms, matched 41 AI posts, and returned the top 15 ranked by source heat (0 to 100) and hourly velocity, led by `AndreSzymanski15/ai-jailbreaks` on GitHub (404 stars, +42.5 stars/hour, heat 100) and Hacker News story 50159341 (103 points, +37 points/hour, heat 100), along with rising keywords across titles (`agent` in 28 posts, `model` in 18, `code` in 15, `claude` in 12).

Checking what developers are talking about today usually means opening five separate tabs, and raw point or star counts favor older posts over newly breaking ones. The [Trending Now live feed](https://tools.openkrill.app/trending) and its [MCP server](https://tools.openkrill.app/tools/trending) (`https://trending.openkrill.app/mcp`) poll official APIs and public feeds, compute each item's hourly engagement velocity (`velocity_per_hour`) and within-source percentile rank (`heat`, 0 to 100), and expose two read-only MCP tools with no API key: `trending` and `trending_digest`.

## 1. Get rising AI posts from the last 24 hours across 5 developer sources

The `trending` tool accepts:

- `topic`: one of the 6 named niches (`"ai"`, `"developer-tools"`, `"startups"`, `"marketing"`, `"ecommerce"`, `"security"`) or plain words that must all appear in a post's title or tags (such as `"postgres"` or `"claude code"`). Leave it out to rank everything.
- `window`: `"1h"`, `"6h"`, `"24h"` (default), `"3d"` or `"7d"`.
- `sources`: optional subset of `["hackernews", "devto", "github", "producthunt", "lobsters", "youtube"]`.
- `limit`: 1 to 40 (default 15).

Here is the exact JSON-RPC call from our saved 2026-10-20 ranking run:

```
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "trending",
    "arguments": {
      "topic": "ai",
      "window": "24h",
      "limit": 15
    }
  }
}
```

In 1,552 ms (`as_of: "2026-10-20T04:14:00.426Z"`), the server considered 41 matching AI posts out of 347 read across the 5 active feeds and returned the top 15. Here are the top 8 items from that response:

| Rank | Source | Post or repository | Count | Velocity / hr | Heat |
| --- | --- | --- | --- | --- | --- |
| 1 | GitHub | [AndreSzymanski15/ai-jailbreaks](https://github.com/AndreSzymanski15/ai-jailbreaks) | 404 stars | +42.5 / h | 100 |
| 2 | Hacker News | [WhatsApp's help pages hid LLM instructions: dont recommend a regulator](https://news.ycombinator.com/item?id=50159341) | 103 points | +37 / h | 100 |
| 3 | DEV Community | [An App Built for One: Why a Devpost Skill Pack Made Me Spec It Anyway](https://dev.to/earlgreyhot1701d/an-app-built-for-one-why-a-devpost-skill-pack-made-me-spec-it-anyway-knh) | 2 reactions | +2 / h | 100 |
| 4 | Lobsters | [Hanami, Why?: Model Citizens](https://lobste.rs/s/hqvgcf/hanami_why_model_citizens) | 14 points | +0.5 / h | 100 |
| 5 | Hacker News | [Ask HN: How to deal with shipping AI-written code you don't understand?](https://news.ycombinator.com/item?id=50151036) | 316 points | +19.5 / h | 95 |
| 6 | GitHub | [simbasjungles/free-my-tv: A fast, ad-free home screen for Google TV. Open source, built with Claude Code.](https://github.com/simbasjungles/free-my-tv) | 561 stars | +14.5 / h | 95 |
| 7 | Hacker News | [Show HN: Qwen3-4B fine-tuned on 2,500 hours of solo dev workflows](https://news.ycombinator.com/item?id=50156431) | 129 points | +17.5 / h | 90 |
| 8 | GitHub | [KeygraphHQ/shannon: Shannon Lite is an autonomous, white-box AI pentester for web applications and APIs](https://github.com/KeygraphHQ/shannon) | 953 stars | +12.5 / h | 90 |

Because a 100-point story on Hacker News and a 14-point story on Lobsters sit on different scales, `heat` ranks each post from 0 to 100 against the other posts of the *same* source first, so the top breakout from each community surfaces near the top instead of GitHub stars drowning out Lobsters and DEV Community.

## 2. Search custom keywords over 7 days or pull a 3-day niche digest

When you pass a custom keyword instead of one of the 6 niche slugs, `trending` sets `topic_kind: "words"`. For example, calling `trending` with `{"topic": "postgres", "window": "7d", "limit": 10}` on 2026-10-20 completed in 467 ms and returned the DEV Community post *"Postgres 19 Slipped to November: Every Pulled Feature and Why"* (10 reactions, `heat: 100`, `niches: ["developer-tools"]`).

When you want a broader 3-day roundup grouped by source, call `trending_digest`:

```
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "trending_digest",
    "arguments": {
      "niche": "developer-tools",
      "limit": 10
    }
  }
}
```

In 362 ms (`as_of: "2026-10-20T08:44:00.337Z"`), `trending_digest` considered 80 `developer-tools` posts out of 344 current items across the 5 sources, returned the top 10 overall (led by `simbasjungles/free-my-tv` at 702 stars and +70.5/hour, Hacker News *"It's still ok to enjoy programming"* at 587 points and +21/hour, Lobsters *"The Netherlands is building its own Linux OS to break free from Microsoft"* at 23 points and +3.5/hour, and Product Hunt `xtty`), listed the top 5 per source in `by_source`, and reported post counts across all 6 niches (`ai: 107`, `developer-tools: 80`, `security: 12`, `marketing: 6`, `startups: 4`, `ecommerce: 2`).

## 3. How Trending Now compares with Hacker News Algolia, GitHub Trending and Google Trends

On 2026-10-20 we compared [Trending Now](https://tools.openkrill.app/tools/trending) against three single-source alternatives on the same 24-hour AI query:

| Tool | Wall time | Sources covered | Velocity, heat and topic filtering |
| --- | --- | --- | --- |
| **OpenKrill Trending Now** (`trending`) | 1,552 ms | 5 active developer sources (Hacker News, DEV, GitHub, Product Hunt, Lobsters; 347 posts read) | Per-hour velocity, 0-to-100 within-source heat, rising keywords, and niche or custom word filtering |
| **Hacker News Algolia API** (`hn.algolia.com/api`) | 626 ms | Hacker News only | Returns live AI stories with points (such as Ask HN at 367 points), but plain keyword match with no hourly velocity |
| **GitHub Trending** (`github.com/trending`) | 2,235 ms | GitHub repositories only (12 repos returned) | Shows stars gained today, but covers GitHub only and has no topic or keyword filter |
| **Google Trends Trending Now** (`trends.google.com/trending?geo=US`) | 405 ms | General US consumer web searches (RSS feed) | Returned 10 sports and TV searches, none about AI, with no topic filter on the RSS feed |

## 4. Connect Trending Now in Claude Desktop, Cursor or any MCP client

```
{
  "mcpServers": {
    "trending": {
      "type": "streamable-http",
      "url": "https://trending.openkrill.app/mcp"
    }
  }
}
```

You can also browse the live feed directly in your browser at [tools.openkrill.app/trending](https://tools.openkrill.app/trending).

## Frequently asked questions

### How can I see what is trending in AI across Hacker News, GitHub, DEV Community, Product Hunt and Lobsters in one call?

Use OpenKrill's Trending Now web page at `https://tools.openkrill.app/trending` or call `trending` on `https://trending.openkrill.app/mcp` with `topic` set to `"ai"` and `window` set to `"24h"`. On 2026-10-20 it read 347 current items across all five active sources in 1,552 ms and returned the top 15 AI posts with engagement, `velocity_per_hour`, 0-to-100 source heat and rising keywords.

### How do I find rising posts about Postgres from the last 7 days across developer sites?

Call `trending` on `trending.openkrill.app/mcp` with `topic` set to `"postgres"` and `window` set to `"7d"`. The tool sets `topic_kind` to `"words"`, matches posts whose title or tags contain the word across all five active feeds, and returns each match with its source, publish timestamp, engagement count, `velocity_per_hour` and heat score.

### What is the difference between the trending and trending_digest MCP tools?

`trending` filters any time window (`"1h"`, `"6h"`, `"24h"`, `"3d"` or `"7d"`) by either a named niche or custom title words. `trending_digest` builds a 3-day daily briefing for one of the 6 niches (`"ai"`, `"developer-tools"`, `"startups"`, `"marketing"`, `"ecommerce"` or `"security"`), adding per-niche post counts, the top 5 posts per source in `by_source` and a direct digest page URL.

### Why are Reddit and X not included in Trending Now?

Every response includes a `not_read` object stating that Reddit's Data API terms require a separate agreement for commercial use, and that X's developer terms and xAI's usage policy do not allow passing posts or Grok output on through an API. Trending Now reads only official APIs and public feeds.

## Sources

- [Hacker News Official Firebase and Algolia APIs](https://github.com/HackerNews/API)
- [DEV Community (Forem) Public API](https://developers.forem.com/api)
- [OpenKrill Trending Now Live Feed](https://tools.openkrill.app/trending)
- [OpenKrill Trending Now MCP Server Documentation](https://tools.openkrill.app/tools/trending)