Skill v1.0.0

currentAutomated scan100/100

lispking/ferris-search/ferris-search-setup

──Details

PublishedNovember 8, 2026 at 11:22 PM

Content Hashsha256:d4e37c92a30a0716...

Git SHA88d94f8e9166

──Files

Files (1 file, 3.8 KB)

SKILL.md3.8 KBactive

SKILL.md · 108 lines · 3.8 KB

name: ferris-search-setup description: | CRITICAL: Use for ferris-search installation and configuration. Triggers on: ferris-search install, mcp add ferris-search, setup ferris-search, DEFAULT_SEARCH_ENGINE, ALLOWED_SEARCH_ENGINES, BRAVE_API_KEY, EXA_API_KEY, FIRECRAWL_API_KEY, JINA_API_KEY, TAVILY_API_KEY, USE_PROXY, ferris-search config, ferris-search build, ferris-search Docker, 安装ferris-search, 配置搜索引擎, MCP服务器配置, 代理设置 metadata: version: "1.0.0"

ferris-search Setup & Configuration Skill

Version: ferris-search 0.1.0 | Last Updated: 2026-03-30

You are an expert at installing and configuring the ferris-search MCP server. Help users by:

Setup: Guide through build, install, and MCP registration
Configuration: Explain env vars and their effects

Documentation

Refer to the local files for detailed documentation:

./references/configuration.md - All environment variables and their effects

IMPORTANT: Documentation Completeness Check

Before answering questions, Claude MUST:

Read ./references/configuration.md
If file read fails: still answer based on SKILL.md patterns

Key Patterns

Build & register with Claude Code

bash

cargo build --release
claude mcp add ferris-search ./target/release/ferris-search

With environment variables

bash

claude mcp add ferris-search ./target/release/ferris-search \
  -e DEFAULT_SEARCH_ENGINE=bing \
  -e ALLOWED_SEARCH_ENGINES=bing,duckduckgo,brave

Claude Desktop / Cursor (mcp-config.json)

json

{
  "mcpServers": {
    "ferris-search": {
      "command": "/path/to/ferris-search",
      "env": {
        "DEFAULT_SEARCH_ENGINE": "bing",
        "ALLOWED_SEARCH_ENGINES": "bing,duckduckgo,brave,baidu",
        "EXA_API_KEY": "your-key-here"
      }
    }
  }
}

With proxy

bash

claude mcp add ferris-search ./target/release/ferris-search \
  -e USE_PROXY=true \
  -e PROXY_URL=http://127.0.0.1:7890

Docker

bash

docker build -t ferris-search .
docker run -e DEFAULT_SEARCH_ENGINE=bing ferris-search

Configuration Reference

Env Var	Default	Description
`DEFAULT_SEARCH_ENGINE`	`bing`	Engine used when `engines` param is omitted
`ALLOWED_SEARCH_ENGINES`	all 14 engines	Comma-separated allow-list
`BRAVE_API_KEY`	—	Required only for `brave` engine
`EXA_API_KEY`	—	Required only for `exa` engine
`FIRECRAWL_API_KEY`	—	Required only for `firecrawl` engine
`JINA_API_KEY`	—	Required only for `jina` engine
`TAVILY_API_KEY`	—	Required only for `tavily` engine
`GITHUB_TOKEN`	—	Optional for `github`/`github_code` engines (60→5000 req/hr)
`USE_PROXY`	`false`	Enable HTTP/SOCKS5 proxy
`PROXY_URL`	`http://127.0.0.1:7890`	Proxy address
`ENABLE_HTTP_SERVER`	`false`	Enable HTTP/SSE transport alongside stdio
`MODE`	`stdio`	Transport mode: `stdio`, `http`, or `both`
`RUST_LOG`	`info`	Log level: `debug`, `info`, `warn`, `error`

When Writing Code

Always build with --release for production use (~8 MB binary, <10 ms startup)
Set ALLOWED_SEARCH_ENGINES to only the engines you need — reduces attack surface
Never commit EXA_API_KEY, FIRECRAWL_API_KEY, JINA_API_KEY, TAVILY_API_KEY, BRAVE_API_KEY, or GITHUB_TOKEN to source control — use env var injection
For Chinese content workflows, include baidu, csdn, juejin, zhihu in allow-list

When Answering Questions

claude mcp add is the recommended path for Claude Code users
JSON config is needed for Claude Desktop / Cursor
Proxy support works for all engines including those behind GFW
ALLOWED_SEARCH_ENGINES acts as an allow-list — engines not listed are silently filtered out

All versions