hybrid-deep-search_skill

This skill dynamically routes queries between Brave API and OpenAI Codex to balance cost and depth of search.

Python

2.5k

GitHub Stars

6

Bundled Files

2 months ago

Catalog Refreshed

3 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill openclaw/skills --skill hybrid-deep-search

_meta.json291 B
README.md6.0 KB
SKILL_CN.md5.1 KB
SKILL.md6.4 KB
test_examples.sh681 B
test_report.md3.1 KB

Overview

This skill implements Hybrid Deep Search: an intelligent router that auto-selects between a fast, free Brave web search and a deep-analysis OpenAI Codex path. It balances cost and quality by analyzing query complexity and routing to the optimal backend. Manual modes are available to force quick or codex searches when needed.

How this skill works

The router inspects each query using simple NLP rules (keyword matching, length, question patterns, technical terms) to compute a complexity score. Low-score queries go to the Brave API for fast, free results; high-score queries go to OpenAI Codex for synthesis and deeper reasoning. Users can override routing with manual modes, set focus (web, academic, news, youtube), and control results, format, and verbosity.

When to use it

Quick factual lookups where speed and zero cost matter (definitions, dates, simple facts).
News monitoring and single-topic web searches that need fast returns.
Comparisons, multi-source synthesis, and deep analytical tasks requiring reasoning or summarization.
Academic literature searches and domain-focused research requiring context and inference.
Batch or scripted queries where automated routing reduces manual decision-making.

Best practices

Use auto mode as the default to let the router optimize cost vs quality.
Provide clear, specific queries for better routing and results; longer queries with analysis keywords tend toward codex.
Set appropriate max-results to control API usage and response size.
Enable verbose only when debugging routing decisions or inspecting complexity scores.
Keep OpenAI credentials secured in environment variables and enable codex only when deep analysis is required.

Example use cases

Fast fact: python3 scripts/deep_search.py "who created Python?" → quick Brave search, immediate free result.
Deep comparison: python3 scripts/deep_search.py "compare LangChain vs LlamaIndex" --mode auto → routed to Codex for multi-source analysis.
Academic focus: python3 scripts/deep_search.py "recent papers on multi-agent systems" --mode codex --focus academic → targeted literature synthesis.
Batch processing: run multiple queries from a file in auto mode to reduce manual routing decisions and costs.
News monitoring: python3 scripts/deep_search.py "latest AI news" --mode quick --focus news for fast updates.

FAQ

A complexity score is computed from keywords, query length, question patterns, and technical terms; low scores go to Brave (quick), higher scores go to Codex (deep).

Do I need an OpenAI key to use this skill?

No—Brave web search works without extra keys. OpenAI Codex requires an OPENAI_API_KEY for deep searches and paid usage.