A multi-tier AI search routing skill that automates fact-checking and deep research using Perplexity models.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install deep-search
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install deep-search using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Deep Search?
Deep Search is a technical search routing skill designed to provide AI agents with precise, web-informed answers. By leveraging the Perplexity API, it offers a three-tier approach to querying: quick facts, standard research comparisons, and deep analytical reasoning. This allows developers to balance speed and depth depending on the complexity of the query.
As a core part of the Openclaw Skills ecosystem, it provides agents with high-fidelity data from focused sources like academic papers, news, YouTube, and Reddit. The integration of Langfuse observability ensures that every search operation is traceable, making it ideal for developers building transparent and reliable AI workflows.
Deep Search Use Cases
- Retrieving quick facts and simple lookups using the sonar model
- Conducting standard research, product comparisons, and technical how-to guides
- Performing deep market analysis and complex problem-solving via sonar-reasoning-pro
- Monitoring real-time news, academic databases, or social platforms like Reddit and YouTube
How Deep Search Works
- The user or agent initiates a search by specifying a query and a processing tier: quick, pro, or deep.
- The skill routes the request to the corresponding Perplexity model optimized for that specific depth.
- Optional focus filters are applied to narrow the search scope to specific domains such as academic journals or social media.
- The results are returned as structured text, with automatic tracing enabled for developer oversight.
Deep Search Setup
To integrate this skill, ensure you have Python 3.10+ installed and configure your environment variables for API access.
# Set required Perplexity API key
export PERPLEXITY_API_KEY="your_perplexity_key"
# Optional: Set OpenRouter key for tracing
export OPENROUTER_API_KEY="your_openrouter_key"
Deep Search Data Schema & Taxonomy
| Parameter | Option | Description |
|---|---|---|
| tier | quick | Fast lookups (~2s) using sonar |
| tier | pro | In-depth research (~5-8s) using sonar-pro |
| tier | deep | Deep analysis (~10-20s) using sonar-reasoning-pro |
| focus | academic, news, youtube, reddit | Specific data source filtering |
Deep Search Advanced Features
- Focus modes for specialized sourcing including reddit, youtube, and news
- Built-in Langfuse observability tracing for debugging and performance monitoring
- Automatic model routing to optimize for both latency and depth within Openclaw Skills
- Comprehensive support for complex queries requiring reasoning-pro models
SKILL.md
Loading
Related Openclaw Skills
Deep Search
A sophisticated multi-tier Perplexity search skill providing AI-synthesized answers with citations and observability tracing.
Datadog
A streamlined CLI tool for Openclaw Skills to manage Datadog infrastructure, metrics, and incidents via the REST API.
Cost Tracker
A dedicated AI spending monitor for tracking, reporting, and optimizing OpenRouter costs through a powerful command-line interface.
Copper CRM
A zero-dependency Python tool for AI agents to manage relationship intelligence and sales workflows via the Copper CRM API.
Discord Bot
A zero-dependency Python integration for AI agents to manage Discord servers, roles, and messaging via the Discord REST API.
DocuSign e-Signatures
A lightweight Python-based skill for managing DocuSign envelopes, templates, and recipients through a streamlined CLI.
Featured*
