agenttimes
# Agent Times — Live Context Layer for AI Agents
## TL;DR
```bash
# News search:
curl -s "https://agenttimes.live/ask?q=why+is+NVDA+down+today"
# Weather:
curl -s "https://agenttimes.live/ask?q=weather+tokyo"
# Crypto price:
curl -s "https://agenttimes.live/ask?q=bitcoin+price"
# Trending stories:
curl -s "https://agenttimes.live/trending"
```
Response is JSON. `/ask` auto-detects intent: news queries return articles with `title`, `url`, `summary`, `sentiment`, `entities`; `$TICKER` queries search by entity; weather queries return forecasts; crypto queries return prices. Check the `source` field. No API key needed.
## Common Tasks — Copy-Paste Commands
**Ticker search (use $ prefix):**
```bash
curl -s "https://agenttimes.live/ask?q=%24NVDA"
curl -s "https://agenttimes.live/ask?q=%24SPY"
curl -s "https://agenttimes.live/ask?q=%24AAPL"
```
**Weather (any city):**
```bash
curl -s "https://agenttimes.live/ask?q=weather+tokyo"
curl -s "https://agenttimes.live/ask?q=weather+london"
```
**Crypto prices:**
```bash
curl -s "https://agenttimes.live/ask?q=bitcoin+price"
curl -s "https://agenttimes.live/prices?symbol=ETH,SOL,BTC"
```
**News search:**
```bash
curl -s "https://agenttimes.live/ask?q=why+did+NVDA+drop+today"
```
**Job search (remote devops/SRE):**
```bash
curl -s "https://agenttimes.live/news?category=business&tag=jobs&limit=20"
```
**Country-specific news:**
```bash
curl -s "https://agenttimes.live/ask?q=singapore+news"
```
**Filter by any tag:** Add `&tag=TAG` to `/news`. Example tags: `jobs`, `gaming`, `space`, `bitcoin`, `asia`, `europe`, `startups`, `devtools`.
---
## What This Does
Agent Times is a one-stop information API for AI agents. **3,576 RSS feeds** across **14 categories**, **weather forecasts**, and **crypto prices** — all from one `/ask` endpoint.
- **Semantic search + neural re-ranking** — AI understands meaning, then a cross-encoder re-ranks results by true relevance. "Why did NVDA drop" finds "NVIDIA Falls 3% as Investors Weigh GTC Optimism Against Risks"
- **FinBERT sentiment** — finance articles scored by a financial-domain neural network. Correctly identifies "Rally fades as investors take profits" as bearish. Only runs on business, crypto, energy, ai, tech categories for accuracy.
- **Source credibility** — tier 1 (Reuters, BBC) get 1.3x ranking boost, tier 3 (unknown) get 0.8x penalty
- **GLiNER entity extraction** — zero-shot NER detects people, companies, locations, organizations, products, events. Plus ticker/crypto dictionary matching (300+ stocks, 100+ crypto).
- **ModernBERT classification** — each article classified by a fine-tuned ModernBERT model (93.9% accuracy, 149M params) reading actual text, not just RSS feed label. Tags auto-assigned by topic + country.
- **Weather forecasts** — global weather via Open-Meteo. Just ask "weather tokyo" or "weather london"
- **Crypto prices** — real-time prices from Pyth Network oracle (BTC, ETH, SOL, and 10+ more). Just ask "bitcoin price"
- **Country tags** — filter by 60+ countries and regions
- **Webhook subscriptions** — get notified when articles matching your topic arrive
- **4-tier fallback** — semantic search → keyword search → web search (70+ engines) → query suggestions
228,000+ articles with 30-day retention. 6 AI models (embedder on GPU, classifier on GPU, cross-encoder, FinBERT, GLiNER NER, tag centroids).
## When to Use Which Endpoint
| You want to... | Use this | Free? |
|---|---|---|
| Search for a topic | `GET /ask?q=TOPIC` | Yes |
| Check weather | `GET /ask?q=weather+tokyo` or `GET /weather?location=tokyo` | Yes |
| Get crypto prices | `GET /ask?q=bitcoin+price` or `GET /prices?symbol=BTC` | Yes |
| Get trending stories | `GET /trending` | Yes |
| Browse a category | `GET /news?category=CAT` | Yes |
| Filter by country/tag | `GET /news?category=world&tag=singapore` | Yes |
| Search the web | `GET /search?q=QUERY` | $0.001 (x402) |
| Monitor a topic | `POST /subscribe` | Yes |
| List categories | `GET /news/categories` | Yes |
**Rule of thumb:** Start with `/ask`. It's free, smart, and auto-detects if you're asking about news, weather, or crypto prices. Use `/news` only when you need bulk category browsing with filters.
## /ask — Smart Search (FREE)
Searches all 228K+ articles using AI semantic matching + keyword matching + web search fallback. Returns results ranked by relevance and recency.
```bash
curl -s "https://agenttimes.live/ask?q=bitcoin+etf"
curl -s "https://agenttimes.live/ask?q=singapore+startup+funding"
curl -s "https://agenttimes.live/ask?q=climate+policy+europe&limit=20"
curl -s "https://agenttimes.live/ask?q=remote+devops+jobs"
```
| Param | Default | Description |
|-------|---------|-------------|
| q | required | Search query. Replace spaces with `+` |
| limit | 10 | Max results (1-50) |
**Response:**
```json
{
"success": true,
"query": "bitcoin etf",
"source": "news_db",
"count": 10,
"results": [
{
"title": "SEC Approves Spot Bitcoin ETF Applications",
"url": "https://example.com/article",
"summary": "The Securities and Exchange Commission has approved...",
"category": "crypto",
"published": "Mon, 24 Mar 2026 14:30:00 +0000",
"sentiment": "bullish",
"sentiment_score": 0.75,
"credibility": "high",
"entities": {
"companies": ["SEC"],
"tickers": [],
"crypto": ["BTC"],
"people": []
},
"tags": ["markets"]
}
]
}
```
When `/ask` finds fewer than 3 results, it automatically falls back to web search. When 0 results, it returns `suggestions` with related categories and terms.
## /trending — Trending Stories (FREE)
Detects stories being covered by multiple independent sources.
```bash
curl -s "https://agenttimes.live/trending?hours=12"
```
| Param | Default | Description |
|-------|---------|-------------|
| hours | 6 | Lookback window (1-48) |
| min_sources | 3 | Minimum sources covering the same story |
| limit | 20 | Max trending clusters |
## /subscribe — Topic Alerts (FREE)
Get webhook notifications when new articles matching your query arrive.
```bash
curl -s -X POST https://agenttimes.live/subscribe \
-H "Content-Type: application/json" \
-d '{"query":"bitcoin regulation","category":"crypto","webhook":"https://your-agent.com/notify"}'
```
Response: `{"success":true,"subscription_id":1,"secret":"abc123...","note":"Save this secret — you need it to unsubscribe."}`
Unsubscribe (requires the secret from subscribe response):
```bash
curl -s -X DELETE "https://agenttimes.live/subscribe/1?secret=abc123..."
```
Your webhook receives a POST with matching articles every time new ones arrive (checked every 5 minutes).
**Security:** Webhook must be a public HTTPS URL. Localhost, private IPs, and cloud metadata endpoints are blocked.
## /news — Category Browsing (FREE)
Browse and filter articles by category with optional tag, date range, and keyword filters.
```bash
# All crypto news
curl -s "https://agenttimes.live/news?category=crypto&limit=10"
# Asian news only
curl -s "https://agenttimes.live/news?category=world&tag=asia"
# Singapore news
curl -s "https://agenttimes.live/news?category=world&tag=singapore"
# Gaming news
curl -s "https://agenttimes.live/news?category=entertainment&tag=gaming"
# Jobs in tech
curl -s "https://agenttimes.live/news?category=business&tag=jobs"
# AI news from last 2 days
curl -s "https://agenttimes.live/news?category=ai&since=2026-03-22"
# Search within a category
curl -s "https://agenttimes.live/news?category=tech&q=rust+programming"
```
| Param | Default | Description |
|-------|---------|-------------|
| category | required | One of 14 categories, or `all` |
| tag | — | Subcategory/country filter (see table below) |
| limit | 20 | Max results (1-1000) |
| q | — | Keyword filter within category |
| since | — | ISO date — articles after this time |
| before | — | ISO date — articles before this time |
| dedup | true | Deduplicate similar headlines |
## /search — Web Search ($0.001)
Search the open web via 70+ engines when you need non-news results.
```bash
curl -s "https://agenttimes.live/search?q=how+to+deploy+docker"
```
| Param | Default | Description |
|-------|---------|-------------|
| q | required | Search query |
| limit | 5 | Max results (1-50) |
| category | general | general, news, images, videos, science, files |
## Categories + Tags
14 categories. Use `&tag=` on `/news` to filter by subcategory, country, or region.
| Category | What's in it | Available tags |
|----------|-------------|----------------|
| **world** | International news, geopolitics, diplomacy | `asia`, `europe`, `middleeast`, `africa`, `india`, `latam`, `oceania`, `defense`, `government` + 50 country tags (see below) |
| **politics** | Elections, policy, legislation, regulation | — |
| **business** | Finance, commerce, employment, markets | `finance`, `startups`, `jobs`, `marketing`, `ecommerce`, `fintech`, `realestate`, `legal`, `supplychain` |
| **tech** | Software, hardware, engineering, data | `devtools`, `engineering`, `mobile`, `datascience`, `telecom` |
| **ai** | Artificial intelligence, machine learning | `robotics`, `research` |
| **crypto** | Cryptocurrency, blockchain, DeFi | `web3`, `defi`, `bitcoin`, `markets` |
| **science** | Research, discoveries, academia | `biotech`, `space`, `research` |
| **health** | Medicine, wellness, public health | `fitness` |
| **energy** | Energy, climate change, environment | `climate`, `environment`, `agriculture` |
| **security** | Cybersecurity, infosec, threats | — |
| **sports** | All sports coverage | — |
| **entertainment** | Film, music, gaming, TV, culture | `gaming`, `film`, `music` |
| **lifestyle** | Food, travel, fashion, daily life | `food`, `travel`, `fashion`, `design`, `education` |
| **automotive** | Vehicles, EVs, transportation, logistics | `shipping` |
### Country Tags (use with `category=world&tag=COUNTRY`)
**Asia:** `singapore`, `malaysia`, `philippines`, `indonesia`, `thailand`, `vietnam`, `japan`, `korea`, `china`, `taiwan`, `hongkong`, `india`, `pakistan`, `bangladesh`
**Europe:** `uk`, `germany`, `france`, `italy`, `spain`, `netherlands`, `sweden`, `norway`, `denmark`, `finland`, `ireland`, `poland`, `austria`, `switzerland`, `iceland`, `romania`, `hungary`, `greece`, `serbia`, `ukraine`, `russia`
**Middle East:** `israel`, `turkey`, `iran`, `saudiarabia`, `uae`, `qatar`
**Africa:** `nigeria`, `southafrica`, `kenya`, `ghana`, `ethiopia`
**Latin America:** `brazil`, `mexico`, `argentina`, `colombia`, `chile`, `peru`
**Oceania:** `australia`, `newzealand`
**North America:** `usa`, `canada`
## Response Format
**News results** (`/ask`, `/news`, `/trending`) return:
```json
{"success": true, "count": N, "results": [...]}
```
**Weather** (`/ask?q=weather+tokyo`) returns: `{"source": "open-meteo", "location": "...", "current": {...}, "forecast": [...]}`
**Crypto prices** (`/ask?q=bitcoin+price`) returns: `{"source": "pyth", "prices": [{"symbol": "BTC", "price": 66000}]}`
Check the `source` field to determine response shape: `news_db`, `combined`, `open-meteo`, `nea`, or `pyth`.
Each article result contains:
- `title` — headline
- `url` — link to original article
- `summary` — first 200 chars of content
- `category` — one of 14 categories
- `source_feed` — RSS feed URL
- `published` — publication date
- `timestamp` — Unix timestamp (ms)
- `sentiment` — "bullish", "bearish", or "neutral"
- `sentiment_score` — -1.0 (very bearish) to 1.0 (very bullish)
- `credibility` — "high" (tier 1 outlets), "medium" (established), or "unknown"
- `entities` — `{people, companies, tickers, crypto, locations}` extracted by GLiNER + dictionary matching
- `tags` — subcategory and country tags for this article
Errors: `{"success": false, "error": "Missing param: q"}`
## Payment (x402)
The `/search` endpoint uses the [x402](https://www.x402.org) protocol. Your agent pays **$0.001 USDC** per request on Base network — no accounts, no subscriptions, just micropayments. All other endpoints are free.
Wallet: `0x536Eafe011786599d9a656D62e2aeAFcE06a96D1` (Base)
## Other Free Endpoints
| Endpoint | Description |
|----------|-------------|
| `GET /news/categories` | List all 14 categories with counts |
| `GET /stats` | Database statistics |
| `GET /feeds/health?category=CAT` | Per-feed health and status |
| `GET /health` | Health check |
| `GET /info` | Full API docs (JSON for agents, HTML for browsers) |
## Tips for Best Results
1. **Always start with `/ask`** — it's free, semantic, and handles fallback automatically
2. **Country news:** Use `/ask?q=japan+earthquake` not `/news?category=japan` (no country categories)
3. **Filter by country:** Use `/news?category=world&tag=singapore` for source-based filtering
4. **Sentiment analysis:** Check `sentiment_score` to gauge market/public mood on a topic
5. **Entity detection:** Use `entities.tickers` and `entities.crypto` to find articles mentioning specific assets
6. **Monitor breaking news:** POST to `/subscribe` with a webhook URL to get real-time alerts
7. **Parse JSON directly** — don't pipe to `jq` (may not be installed)
8. **Replace spaces with `+`** in all query strings
9. **Check `suggestions`** when 0 results — it recommends related categories and search terms
10. **Credibility matters:** Sort by `credibility` to prioritize tier 1 sources (Reuters, BBC, AP) over unknown blogs
标签
skill
ai