How-To Series · Episode 21 / 59 · Module 4: Eyes, Ears, Voice

Hermes · Web Search

Two tools · eight backends · smart about big pages. Pick by cost and feature.

After this videoYou can now wire the web into Hermes the way that fits your budget.

Two tools: web_search ranks results, web_extract pulls readable content (with deep crawl when the backend supports it). Eight backends — Firecrawl (default), SearXNG (free self-hosted), Brave, DDGS (DuckDuckGo, no key), Tavily, Exa, Parallel, xAI Grok. Mix per capability: free search + paid extract is a two-line config. web_extract is size-aware: under 5K chars returns raw, 5K-500K summarizes via an auxiliary model, 500K-2M chunks and synthesizes, over 2M refuses. Route the web_extract aux model to a cheap fast one (Gemini Flash) so summarizing long pages does not burn Opus tokens.

About these resources. Every command comes from the Web Search doc; the Auxiliary Models reference is cited for the cost knob.

Sources · What this video distills

2 docs pages · every command below traces to one of them
Primary · backends, per-capability split, size-driven extract pipeline
Web Search & Extract
Read ↗
The auxiliary.web_extract slot in context
Auxiliary Models
Read ↗

Commands shown · Copy and paste

each shows the source doc it came from
Portal pathfrom source ↗
hermes setup --portal && hermes tools
echo "FIRECRAWL_API_KEY=fc-***" >> ~/.hermes/.env
Cheap aux model (config.yaml)from source ↗
auxiliary.web_extract: { provider: openrouter, model: google/gemini-3-flash-preview }

Going deeper · Related Hermes docs

further reading · not sources of facts shown above

Next in the series · Episodes that build on this

E20
Browser Automation
E11
Tools & Toolsets
E53
Tool Gateway