modsearch

Adoption judgment: modsearch is worth trying if you are adding web retrieval to Claude Code, Codex, Cursor, or another Agent Skill, but the current evidence supports experimentation by technical users rather than treating it as mature search infrastructure. The GitHub repository describes a CLI toolkit that turns queries into structured web evidence JSON and uses an extensible provider architecture intended for Agent Skill calls. It is not an independent search engine, a full web crawler, a knowledge base, or a report-writing system; a more accurate analogy is a CLI orchestration and adapter layer between an agent, search providers, and web-page extraction.

In practice, it is intended to route a query through configurable providers and return results in a format an agent can filter, cite, or summarize. Project-authored X posts show web search and web parsing demonstrations and mention integrations or routes involving Exa, Firecrawl, Tavily, Antigravity CLI, and X search through Grok Build. These posts establish possible integration paths, but they are demonstrations or social claims rather than stable guarantees for every provider, page type, or result set.