Deep web research with Brave Search API. Find sources, extract content, and compile comprehensive reports. Use when a user asks to research a topic on the web, find information online, compile a research report, search the internet for sources, do a deep dive on a subject, gather web sources for a report, or investigate a topic thoroughly using multiple online sources.
Conduct thorough web research using the Brave Search API. Search for information, fetch and extract content from web pages, cross-reference multiple sources, and compile structured research reports with proper citations. Designed for deep-dive research that goes beyond a single search query.
When a user asks you to research a topic on the web, follow these steps:
Before searching, define a research plan:
Example plan for "Is Rust ready for production web development?":
Central question: Can Rust be used for production web APIs today?
Sub-queries:
1. "Rust web frameworks comparison 2025"
2. "Rust production web services case studies"
3. "Rust vs Go web performance benchmarks"
4. "Rust web development challenges limitations"
5. "companies using Rust in production backend"
import requests
import os
BRAVE_API_KEY = os.environ.get("BRAVE_API_KEY")
BRAVE_SEARCH_URL = "https://api.search.brave.com/res/v1/web/search"
def brave_search(query, count=10, freshness=None):
"""Search the web using Brave Search API.
Args:
query: Search query string
count: Number of results (max 20)
freshness: Optional filter: 'pd' (past day), 'pw' (past week),
'pm' (past month), 'py' (past year)
"""
headers = {
"Accept": "application/json",
"Accept-Encoding": "gzip",
"X-Subscription-Token": BRAVE_API_KEY,
}
params = {"q": query, "count": min(count, 20)}
if freshness:
params["freshness"] = freshness
response = requests.get(BRAVE_SEARCH_URL, headers=headers,
params=params, timeout=30)
response.raise_for_status()
results = []
data = response.json()
for item in data.get("web", {}).get("results", []):
results.append({
"title": item["title"],
"url": item["url"],
"description": item.get("description", ""),
"age": item.get("age", ""),
})
return results
Fetch and extract readable content from the most relevant URLs:
from bs4 import BeautifulSoup
def extract_page_content(url):
"""Fetch a URL and extract the main text content."""
headers = {"User-Agent": "research-bot/1.0.0"}
response = requests.get(url, headers=headers, timeout=30)
response.raise_for_status()
soup = BeautifulSoup(response.text, "html.parser")
# Remove navigation, ads, footers
for tag in soup.select("nav, footer, header, aside, .ad, .sidebar, script, style"):
tag.decompose()
# Try to find the main content area
main = soup.select_one("article, main, .post-content, .entry-content")
if not main:
main = soup.find("body")
if main:
text = main.get_text(separator="\n", strip=True)
# Clean up excessive whitespace
lines = [line.strip() for line in text.split("\n") if line.strip()]
return "\n".join(lines)
return ""
For each sub-question, synthesize information from multiple sources:
Sub-question: "Rust web frameworks comparison"
Source 1 (blog.example.com): Actix-web is the fastest, Axum has the
best developer experience, Rocket is easiest for beginners.
Source 2 (benchmark-site.com): Actix-web handles 650K req/s, Axum
handles 580K req/s. Both outperform Express.js by 10x.
Source 3 (forum discussion): Community consensus is shifting toward
Axum due to its use of the Tokio ecosystem and simpler API.
Synthesis: Axum is emerging as the recommended choice, offering a
balance of performance and ergonomics, with strong ecosystem support.
# Research Report: [Topic]
**Date:** [date]
**Queries executed:** [count]
**Sources analyzed:** [count]
## Executive Summary
[2-3 paragraph summary of key findings]
## Detailed Findings
### [Sub-topic 1]
[Findings with inline citations]
Key data points:
- [fact] [Source 1]
- [fact] [Source 2]
### [Sub-topic 2]
[Findings with inline citations]
## Conflicting Information
[Note any areas where sources disagree and why]
## Knowledge Gaps
[Areas where information was insufficient or outdated]
## Sources
1. [Title](URL) - [brief description of what was extracted]
2. [Title](URL) - [brief description]
3. [Title](URL) - [brief description]
...
## Methodology
- Search engine: Brave Search API
- Queries executed: [list each query]
- Date range: [freshness filter used]
- Sources evaluated: [total count]
- Sources included: [count included in report]
cat > research_report_[topic].md << 'EOF'
[compiled report]
EOF
User request: "Research whether we should migrate from REST to GraphQL for our mobile app API."
Research plan:
Output: Structured report with pros, cons, case studies, and a recommendation based on evidence.
User request: "Research the current state of AI code review tools. Who are the players and what are the gaps?"
Research plan:
Output: Competitive landscape analysis with feature matrices, pricing data, user sentiment, and identified market gaps.
User request: "Research whether the claim 'microservices reduce deployment frequency' is supported by evidence."
Research plan:
Output: Evidence-based analysis citing research papers, industry surveys, and case studies both supporting and refuting the claim.
freshness parameter to prioritize recent results.npx skills add TerminalSkills/web-research下载完整 Skill 目录，包含 SKILL.md 及所有相关文件
Category:web-search