Nimble search returns JSON regardless of output flags

Nimble search returns JSON regardless of output flags

TL;DR

Using the Nimble CLI for web research: set NIMBLEAPIKEY once per session (create the key in the dashboard first) and stop re-authenticating per command. Decide up front: search when the source is unknown, extract when you have the URL. Don't chain search-into-extract blindly; if you already know the page, extract directly.

Use this when you run into the situation in the title.

When not to use this skill: unrelated tasks. It covers only the procedure above.

Compatibility

The steps above apply to the commands named in them. This skill does not pin a version, so if a flag looks different on your machine, check your installed version's docs first.

Details

Keep result handling consistent, the CLI returns JSON you can pipe into jq for URLs and titles.

Context: Web (Nimble official cookbook, LangChain web agent CLI): documents the CLI search workflow. API key creation happens in the Nimble dashboard and is read via NIMBLEAPIKEY. The gotcha for agents: pick the right tool for the job. Use the search tool for open-ended discovery across sources, and the extract tool when you already know the exact page; wrapping one inside the other wastes calls. Configure the CLI once per session rather than re-authenticating per call.

Source: Web (Nimble official cookbook, LangChain web agent CLI): documents the CLI search workflow. API key creation happens in the Nimble dashboard and is read via NIMBLEAPIKEY. The gotcha for agents: pick the right tool for the job. Use the search tool for open-ended discovery across sources, and the extract tool when you already know the exact page; wrapping one inside the other wastes calls. Configure the CLI once per session rather than re-authenticating per call.

Verified resolution: Using the Nimble CLI for web research: set NIMBLEAPIKEY once per session (create the key in the dashboard first) and stop re-authenticating per command. Decide up front: search when the source is unknown, extract when you have the URL. Don't chain search-into-extract blindly; if you already know the page, extract directly. Keep result handling consistent, the CLI returns JSON you can pipe into jq for URLs and titles.

Source URL: https://github.com/nimbleway/cookbook/blob/HEAD/langchain/webagentcli/README.md