Deep Search for Openclaw

A sophisticated multi-tier Perplexity search skill providing AI-synthesized answers with citations and observability tracing.

aiwithabidi
v1.0.0
Feb 15, 2026
0
786
1

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install deep-search-pro

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install deep-search-pro 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 high-performance research skill designed to give your AI agents access to real-time information with varying levels of depth. By leveraging the Perplexity API, it allows agents to choose between quick lookups and deep reasoning tasks, ensuring that information retrieval is both efficient and accurate. Integrating this into your Openclaw Skills ecosystem provides a robust foundation for any task requiring fact-checking, synthesis, or specialized domain research.

The skill stands out by offering dedicated focus modes and built-in observability. Whether you need to pull from academic papers or monitor the latest news, Deep Search provides structured, cited responses that your agents can rely on for complex decision-making and analysis.

Deep Search Use Cases

  • Rapid fact-checking and simple lookups for immediate answers using the quick tier.
  • In-depth product comparisons and technical how-to research via the pro tier.
  • Comprehensive market analysis and complex problem solving using the deep reasoning model.
  • Targeted searches across specialized silos such as academic journals, news outlets, YouTube, or Reddit.

How Deep Search Works

  1. The skill is triggered by specific research-oriented keywords like search, compare, or research.
  2. The system determines the required depth based on the command (quick, pro, or deep).
  3. A query is dispatched to the Perplexity API using the appropriate Sonar model variant.
  4. If configured, Langfuse traces the request to provide observability and monitor model performance.
  5. The skill processes the API response to return a synthesized, citation-backed answer to the agent.

Deep Search Setup

To get started with this addition to your Openclaw Skills, ensure python3 is installed and follow these steps:

# Set your required Perplexity API Key
export PERPLEXITY_API_KEY='your_api_key_here'

# Optional: Set OpenRouter key for enhanced tracing metadata
export OPENROUTER_API_KEY='your_api_key_here'

Configure your agent to point to the deep_search.py script within your local environment.

Deep Search Data Schema & Taxonomy

Component Description
Tiers Three levels of depth: quick (sonar), pro (sonar-pro), and deep (sonar-reasoning-pro).
Focus Modes Customizable search filters including internet, academic, news, youtube, and reddit.
Citations Structured references provided for all synthesized information.
Tracing Data Observability logs sent to Langfuse for debugging and cost management.

Deep Search Advanced Features

  • Dynamic tier selection to optimize for either speed (2s) or deep reasoning (up to 20s).
  • Native Langfuse integration for comprehensive observability and tracing of research workflows.
  • Domain-specific focus modes for high-signal data retrieval from academic and social platforms.
  • Full compatibility with the Openclaw Skills framework for multi-agent research pipelines.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*