A search engine skill that speaks four languages to please your AI agent

What it does

AnySearch is a skill — essentially a discoverable plugin — that gives AI agents real-time web search, batch parallel search, and full-page content extraction. It is a thin client: the actual search happens at anysearch.com, and the repo provides four local CLI wrappers (Python, Node.js, PowerShell, Bash) that all talk to the same API.

The interesting bit

The project treats “what runtime is available” as a configuration problem rather than an assumption. It ships identical CLIs in four languages, includes a probing ritual to detect the cleanest runtime, and persists the winner to runtime.conf so agents do not have to guess every time. That is unusual attention to heterogenous environments for what is otherwise a straightforward API client.

Key highlights

• Supports general web search, vertical domain search, parallel batch search, and full-page extraction to Markdown
• Anonymous access works without an API key; a free key raises rate limits
• SKILL.md provides a formal skill definition for agent platforms (Claude Code, Cursor, OpenClaw, etc.)
• Shared install path at ~/.agents/skills/ lets multiple tools use the same copy
• extract output is Markdown by default; the CLI rejects redundant format flags

Caveats

• The README is almost entirely installation and verification instructions; the actual API surface is barely documented beyond four subcommands
• No source code for the search backend is included — this is strictly a client skill, not a self-hosted engine

Verdict

Worth a look if you are building or curating an AI agent toolchain and need a drop-in search capability with minimal runtime assumptions. Skip it if you wanted an open search index or a standalone library; this is glue code, competently written glue code, but glue code nonetheless.