Skip to content

update exa sample to showcase highlights, max_age_hours, and research paper category #248

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
manoj-selvakumar5 merged 1 commit into from
Apr 22, 2026
+17 −13
Merged

update exa sample to showcase highlights, max_age_hours, and research paper category #248

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 6 additions & 4 deletions python/03-integrate/tools/exa/.prompt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,10 +15,11 @@ Use exa_search with:
- type: "auto" ← SHOWCASES: Intelligent search mode selection
- num_results: 5
- text: {"maxCharacters": 2000} ← SHOWCASES: Content extraction with limits
- highlights: {"maxCharacters": 4000} ← SHOWCASES: Token-efficient page excerpts
- summary: {"query": "What are the key concepts, main points, and important details?"} ← SHOWCASES: AI summaries
- subpages: 2 ← SHOWCASES: Subpage crawling to find related content
- subpage_target: ["overview", "about", "introduction"] ← SHOWCASES: Targeted subpage discovery
- livecrawl: "fallback" ← SHOWCASES: Live crawling when cache misses
- max_age_hours: 168 ← SHOWCASES: Content freshness (livecrawl pages older than 7 days)

PURPOSE: Get foundational understanding with AI-summarized key points.

Expand All @@ -33,7 +34,7 @@ Use exa_search with:
- start_published_date: [Use the date provided in the prompt] ← SHOWCASES: Date filtering
- text: {"maxCharacters": 1500}
- summary: {"query": "What are the key announcements, developments, and news?"}
- livecrawl: "fallback" ← SHOWCASES: Fresh news retrieval
- max_age_hours: 24 ← SHOWCASES: Fresh news (livecrawl pages older than 24h)

PURPOSE: Capture recent developments and breaking news.

Expand All @@ -43,7 +44,7 @@ STEP 3: ACADEMIC PAPERS - Demonstrating PDF Category & Structured Output

Use exa_search with:
- query: Academic/research-focused query (include terms like "research", "study", "analysis")
- category: "pdf" ← SHOWCASES: Academic/PDF document search
- category: "research paper" ← SHOWCASES: Academic/research paper search
- num_results: 5
- text: {"maxCharacters": 2000}
- summary: {
Expand Down Expand Up @@ -82,10 +83,11 @@ STEP 5: DEEP DIVE - Demonstrating exa_get_contents for Detailed Extraction
Select 2-3 of the most valuable URLs from previous searches, then use exa_get_contents with:
- urls: [list of important URLs]
- text: {"maxCharacters": 4000} ← SHOWCASES: Extended content extraction
- highlights: {"maxCharacters": 4000} ← SHOWCASES: Key excerpts from important sources
- summary: {"query": "Extract all important details, insights, and actionable information"}
- subpages: 3 ← SHOWCASES: Deep subpage discovery (citations, references, docs)
- subpage_target: ["references", "citations", "bibliography", "methodology"] ← SHOWCASES: Targeted crawling for academic context
- livecrawl: "always" ← SHOWCASES: Force fresh content retrieval
- max_age_hours: 0 ← SHOWCASES: Force fresh content (always livecrawl)

PURPOSE: Get comprehensive detail from the most important sources, including citations and methodology.

Expand Down
11 changes: 6 additions & 5 deletions python/03-integrate/tools/exa/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,13 +11,14 @@ A conversational AI agent that demonstrates the full power of Exa's search and c
| Feature | Description |
|---------|-------------|
| Auto Mode | Intelligent search mode selection for optimal results |
| Category Filtering | Specialized searches for news, PDF documents, and GitHub repositories |
| Category Filtering | Specialized searches for news, research papers, and GitHub repositories |
| Date Filtering | Time-bound searches for recent content (e.g., last 30 days) |
| AI Summaries | Automatic key insights extraction from search results |
| Structured Output | JSON schema for structured summary extraction |
| Subpage Crawling | Discover related pages (citations, methodology, references) |
| Subpage Targeting | Keywords to find specific subpages (references, bibliography) |
| Live Crawling | Fresh content retrieval, bypassing cache |
| Highlights | Token-efficient page excerpts for key information |
| Content Freshness | max_age_hours for livecrawl control (replaces old livecrawl enum) |
| Content Extraction | Full text retrieval with character limit control |

## Architecture
Expand All @@ -42,9 +43,9 @@ The Deep Research Assistant implements a comprehensive 6-step research workflow:
│ ├─────────────────────────────────────────────────────────────┤ │
│ │ 1. Overview Search │ Auto mode + subpages + AI summaries │ │
│ │ 2. News Search │ Category: news + date filtering │ │
│ │ 3. Academic Papers │ Category: pdf + structured output │ │
│ │ 3. Academic Papers │ Category: research paper + structured │ │
│ │ 4. GitHub Projects │ Category: github │ │
│ │ 5. Deep Dive │ exa_get_contents + live crawling │ │
│ │ 5. Deep Dive │ exa_get_contents + highlights │ │
│ │ 6. Synthesis │ Comprehensive research brief │ │
│ └─────────────────────────────────────────────────────────────┘ │
│ │
Expand Down Expand Up @@ -305,4 +306,4 @@ Store API keys (such as `EXA_API_KEY`) securely using environment variables and

### Clean up

Clean up the resources created on AWS to avoid undesired costs.
Clean up the resources created on AWS to avoid undesired costs.
9 changes: 5 additions & 4 deletions python/03-integrate/tools/exa/deep_research_assistant.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,19 +7,20 @@

Exa Capabilities Demonstrated:
- Auto mode: Intelligent search mode selection for optimal results
- Category filtering: news, pdf, github for specialized searches
- Category filtering: news, research paper, github for specialized searches
- Date filtering: Time-bound searches for recent content
- AI summaries: Automatic key insights extraction
- Highlights: Token-efficient page excerpts
- Structured output: JSON schema for structured summary extraction
- Subpage crawling: Discover related pages (citations, methodology)
- Subpage targeting: Keywords to find specific subpages (references, bibliography)
- Live crawling: Fresh content retrieval, bypassing cache
- Content freshness: max_age_hours for livecrawl control
- Content extraction: Full text with character limit control

Research Workflow:
1. Overview Search (Auto Mode) - General topic understanding
2. News Search (Category: news) - Recent developments with date filtering
3. Academic Papers (Category: pdf) - Technical depth from research
3. Academic Papers (Category: research paper) - Technical depth from research
4. Code & Projects (Category: github) - Practical implementations
5. Deep Dive (exa_get_contents) - Detailed extraction from key sources
6. Synthesis - Structured research brief with citations
Expand Down Expand Up @@ -126,7 +127,7 @@ def main():
print("Deep Research Assistant")
print("=======================")
print("Demonstrating Exa capabilities: auto mode, category filters, date filtering,")
print("AI summaries, structured output (JSON schema), subpage crawling/targeting, and live crawling.\n")
print("AI summaries, highlights, structured output, subpage crawling/targeting, and content freshness.\n")

agent = create_research_agent()

Expand Down
Loading