Ultimate Web Search Skill — AI skill for Claude Code
LLM-neutral web search skill for AI coding agents.
How to install Ultimate Web Search Skill
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open code-yeongyu/ultimate-web-search-skill and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Ultimate Web Search Skill does
LLM-neutral web search skill for AI coding agents. 12 providers behind one CLI - cross-platform, stdlib-only Python, with sequential fallback chain.
Alternatives in AI
- Cumora — Where agent teams gather 3.2k ★
- onWatch — Open-source Go CLI that tracks AI API quota usage across 7 providers (Synthetic, Z.ai, Anthropic, Codex, GitHu 397 ★
- Pi Agent Desktop — Pi — A cross-platform AI coding agent, bringing the Claude Code experience to your desktop 289 ★
README
web-search
LLM-neutral web search skill for AI coding agents (pi, hermes-agent, openclaw, opencode, claude-code, etc.).
Instead of hard-coding a single search provider into every agent, this skill lets the agent invoke a single command (`scripts/web-search`) that fans out across **12 search providers**, dumps raw + normalized results to a temp directory, and prints a path the agent can pipe through `rg`, `jq`, or any other tool.
**Zero config out of the box** — DuckDuckGo and MWMBL accept anonymous calls. Add API keys for richer providers (Tavily, Brave, Perplexity, Anthropic, OpenAI, etc.) when you want them.
Why a skill, not a library?
Agent harnesses already have a way to load skills on demand. Embedding web search as a skill means:
- No fixed context cost — the skill loads only when the agent needs to search.
- No vendor lock-in — swap providers per call, or fan out across many in parallel.
- No SDK churn — pure Python 3 stdlib + curl. Works wherever Python 3 runs.
- Pipeline-native — results live as JSON files on disk so
rg,jq,grep,python -call just work.
Supported providers (12)
| Provider | Auth | Notes |
|---|---|---|
duckduckgo |
none | Free Instant Answer API. Zero-config. |
mwmbl |
none | Free open-source independent index. Zero-config. |
exa |
API key | Semantic / neural search. (Anonymous calls disabled in 2026.) |
tavily |
API key | LLM-friendly result snippets. |
brave |
API key | Independent index. Good for privacy-sensitive queries. |
serper |
API key | Google results via paid proxy. |
google-cse |
API key + Custom Search Engine ID | Official Google. Max 10 results per call. |
z-ai |
API key | China-friendly index. |
perplexity |
API key | Pure search OR sonar with built-in answer. |
xai |
API key | Grok web_search + x_search via Responses API. |
openai |
API key | OpenAI hosted web search (Responses API + web_search tool). |
anthropic |
API key | C |
Related Skills
OmniPilot
Cross-platform LLM computer-use agent — OpenAI, Anthropic, Gemini, NVIDIA NIM and more control your PC behind
Ragent
Agentic RAG chatbot that answers questions from uploaded PDFs using semantic search (FAISS + embeddings), with
Jarhead
Voice-first Mac assistant that uses the computer for you. GPT-Live-1 full-duplex voice behind an on-device wak
Living Codebase Cartographer
Offline codebase cartographer: language-agnostic repo maps, dependency/call graphs, blast-radius impact analys
Cowork
Multi-Agent Cowork MCP Server — cross-platform AI agent coordination with Web UI Dashboard. Ships as Cowork Pr
Flywheel Gateway
SDK-first orchestration platform for managing AI coding agent fleets: BYOA key rotation, real-time WebSocket d
Related Agents
Contract Neutral Reviewer
Contract-neutral fallback reviewer. Executes the attached family review template verbatim when Codex is unavai
Py38 Compat Reviewer
Python 3.8/3.9 互換性のレビューが必要なとき(リリース前、差分レビュー時、CI の 3.8/3.9 ジョブが失敗したとき)に使う。3.10+ でしか動かない構文や 3.9+ の stdlib API を検出
Rmh
Python specialist sub-agent. Principles from Raymond Hettinger's talks, PEPs, and stdlib contributions (collec