Agentic Search

Grok+Tavily/Firecrawl deep research skill with citations and rerankable sessions

Tested · Didn't pass

Cosa fa

A 6-script research toolkit (search, fetch, map, extract, rank, get-sources) that wraps xAI Grok web search plus optional Tavily/Firecrawl for source-backed, citable answers and full-fidelity page-to-Markdown capture. Triggers on requests for current, cited web research or when ordinary search feels thin on sources; not meant for one-off facts.

Rapporto di test

Real run on a clean box: `python3 scripts/agentic_search.py --query test` failed cleanly with `GROK_API_URL is not set` — the skill needs a paid xAI Grok key (plus optional Tavily/Firecrawl keys) and `uv`/Python 3.13 before it produces a single result, so a plain WebSearch beat it 1-0 on this machine.

Testato il: 2026-07-14 · Claude Code 2.x (agent harness)

Installazione

git clone https://github.com/appautomaton/webmaton
cd webmaton
mkdir -p ~/.claude/skills
cp -r skills/agentic-search ~/.claude/skills/agentic-search

Comandi e prompt di esempio

/agentic-searchGrok+Tavily/Firecrawl deep research skill with citations and rerankable sessions

Gli skill si attivano con richieste in linguaggio naturale, senza comandi da ricordare. Dopo l'installazione, prompt come questi lo attivano (in inglese):

Do a deep research pass on prompt caching with real citations
Research this market question and rerank the sources you found
Run a cited literature sweep on agent memory architectures

Rapporto completo →