Griptape Tool: WebSearch

Here's the genuinely nice surprise hiding in the Griptape pack: the WebSearch tool works out of the box with no API key and no account. Its default driver is DuckDuckGo, which is free and doesn't require credentials. That makes this the cheapest way to give your agent real, current web access - which is a bigger deal than it sounds, because most "search the web" solutions in this space want you to sign up for a search API and hand over billing details before they'll do anything.

The catch is the division of labor, same as its sibling WebScraper: WebSearch finds links - it returns search results, not the full content of pages. That's the right tool for "what are the top sources about X?" and the wrong tool for "read me this article." For the latter you pair it with WebScraper, which fetches and reads actual pages. The classic research-agent combo in this pack is both wired into one agent: search to find, scrape to read.

How it works

The node wraps Griptape's WebSearchTool with a search driver and returns it as a TOOL_LIST. When your agent decides a task needs current information, it calls the tool with a query, and the driver returns search results - titles, URLs, snippets - which the agent then reasons over. The default DuckDuckGoWebSearchDriver needs no key; the optional driver input lets you upgrade to a dedicated search API (the pack includes Serper, Tavily, Exa, and Google drivers) if you want more results, better rate limits, or a different ranking.

The inputs that matter

• driver - optional. Leave it empty for the free DuckDuckGo default, or plug in a WEB_SEARCH_DRIVER from the pack's driver nodes for a paid, more reliable engine. The README's video on this node is literally titled "WebSearch Node Now Allows for Driver Functionality," which tells you the driver wiring is a recent, headline feature.
• off_prompt - default False. Keep it on if you want search snippets to stay out of the main model context; the usual cost/privacy trade.

One output, TOOL (TOOL_LIST), feeds an agent.

Install

Standard pack install:

cd ComfyUI/custom_nodes
git clone https://github.com/griptape-ai/ComfyUI-Griptape

or ComfyUI Manager → search "Griptape" → install → restart. No key needed for the default.

Where people get burned

The predictable trap is expecting full page contents back from a search - that's the scraper's job, and without it your agent has links it can't open. Wire in WebScraper too. Second, DuckDuckGo's free tier is what it is: fine for casual use, occasionally rate-limited or thin on results. If you're building something that searches a lot, plan to move to a Serper or Tavily driver (both have free tiers) rather than hammering the default. And prompt explicitly - an agent doesn't search unless it knows it should, so "use your web search to find recent information" beats a vague instruction every time.