Documentation

A search engine for AI tools. Indexes 790+ MCP server tools and 760+ agent skills into a single searchable catalog. No auth required.

GitHub

MCP Server

Native protocol for Claude Code, Cursor, Codex

Agent Skill

Install via npx for any agent

REST API

Simple HTTP endpoints, no SDK needed

MCP Server

Connect Nemo as an MCP server for native tool discovery in your AI coding agent. Once connected, your agent has access to search_tools, get_skill, and call_tool.

claude mcp add --transport http --scope user nemo https://nemo.25chenghua.workers.dev/mcp

Endpoint	https://nemo.25chenghua.workers.dev/mcp
Transport	Streamable HTTP
Auth	None required

Agent Skill

Install Nemo as an agent skill. This adds a SKILL.md to your project that teaches your agent to use Nemo's REST API directly — no MCP server setup needed.

npx skills add ChengHua926/nemo-skill

Compatible with Claude Code, Cursor, Codex, Gemini CLI, and 30+ other agents. The skill teaches your agent to search tools, fetch skill instructions, and call remote MCP tools via simple HTTP requests.

REST API

Use Nemo directly via HTTP. All endpoints return JSON, support CORS, and require no authentication. This is the same API that the agent skill uses under the hood.

GET/api/search

Search tools and skills

GET/api/skill/:name

Get skill instructions

POST/api/call

Call a remote MCP tool

GET/api/search

Search across MCP tools and agent skills. Each result includes a type field: "mcp_tool" or "skill".

Parameters

Name	Type	Default	Description
q	string	required	Search query
limit	number	5	Max results, 1–20
detail	"compact" \| "full"	"compact"	Response detail level
source	"all" \| "mcp" \| "skills"	"all"	Filter by source type

Request

bash

curl "https://nemo.25chenghua.workers.dev/api/search?q=send+email"

Compact response

Default. Returns tool/skill name, server info, and type.

json

[
  {
    "type": "mcp_tool",
    "toolName": "send_email",
    "serverEndpoint": "https://send-email-mcp.example.com/mcp",
    "serverName": "Resend Email",
    "title": null
  },
  {
    "type": "skill",
    "skillName": "email-automation",
    "repo": "acme/agent-skills",
    "description": "Automate email workflows...",
    "installCommand": "npx skills add acme/agent-skills",
    "skillType": "skill"
  }
]

Full response

Set detail=full to include descriptions, input schemas for MCP tools, and full metadata for skills.

bash

curl "https://nemo.25chenghua.workers.dev/api/search?q=send+email&detail=full"

json

[
  {
    "type": "mcp_tool",
    "toolName": "send_email",
    "serverEndpoint": "https://send-email-mcp.example.com/mcp",
    "serverName": "Resend Email",
    "title": null,
    "description": "Send an email via Resend API...",
    "inputSchema": {
      "properties": {
        "sender_email": "string",
        "to_emails": "array",
        "subject": "string"
      },
      "required": ["sender_email", "to_emails", "subject"]
    }
  }
]

Filter by source

bash

curl "https://nemo.25chenghua.workers.dev/api/search?q=deployment&source=skills"

GET/api/skill/:name

Load full instructions for an agent skill. Returns the complete SKILL.md content, install command, and metadata.

Parameters

Name	Type	Default	Description
name	string	required	Skill name (URL path parameter)
repo	string	—	GitHub repo if multiple skills share the same name (query param)

Request

bash

curl "https://nemo.25chenghua.workers.dev/api/skill/vercel-react-best-practices"

Response

json

{
  "skillName": "vercel-react-best-practices",
  "repo": "vercel-labs/agent-skills",
  "description": "React and Next.js performance optimization guidelines...",
  "installCommand": "npx skills add vercel-labs/agent-skills",
  "skillType": "skill",
  "instructions": "# React Best Practices\n\nUse React Server Components by default...",
  "hasScripts": false,
  "hasReferences": true,
  "installCount": 42
}

POST/api/call

Proxy a tool call to any indexed MCP server. Use the serverEndpoint and toolName from search results.

Body parameters

Name	Type	Default	Description
endpoint	string	required	MCP server endpoint URL
tool	string	required	Tool name to call
args	object	{}	Arguments to pass to the tool
maxResponseChars	number	10000	Max response size, 1k–50k chars

Request

bash

curl -X POST "https://nemo.25chenghua.workers.dev/api/call" \
  -H "Content-Type: application/json" \
  -d '{
    "endpoint": "https://send-email-mcp.example.com/mcp",
    "tool": "send_email",
    "args": {
      "sender_email": "you@example.com",
      "to_emails": ["recipient@example.com"],
      "subject": "Hello"
    }
  }'

Response

json

{
  "success": true,
  "content": [
    {
      "type": "text",
      "text": "Email sent successfully to recipient@example.com"
    }
  ]
}

Responses are truncated at maxResponseChars. Increase the value if you need more data.

search_tools

Search across MCP tools and agent skills. Returns compact results by default. Set detail="full" for descriptions and schemas.

Parameters

Name	Type	Default	Description
query	string	required	Search query
limit	number	5	Max results, 1–20
detail	"compact" \| "full"	"compact"	Response detail level
source	"all" \| "mcp" \| "skills"	"all"	Filter by source type

Example

search_tools({ query: "send email" })

[
  {
    "type": "mcp_tool",
    "toolName": "send_email",
    "serverEndpoint": "https://send-email-mcp.example.com/mcp",
    "serverName": "Resend Email",
    "title": null
  },
  {
    "type": "skill",
    "skillName": "email-automation",
    "repo": "acme/agent-skills",
    "description": "Automate email workflows...",
    "installCommand": "npx skills add acme/agent-skills",
    "skillType": "skill"
  }
]

get_skill

Load full instructions for an agent skill found via search_tools. Returns the complete SKILL.md content and install command.

Parameters

Name	Type	Default	Description
name	string	required	Skill name from search results
repo	string	—	GitHub repo if multiple skills share the same name

Example

get_skill({ name: "vercel-react-best-practices" })

{
  "skillName": "vercel-react-best-practices",
  "repo": "vercel-labs/agent-skills",
  "description": "React and Next.js performance optimization guidelines...",
  "installCommand": "npx skills add vercel-labs/agent-skills",
  "instructions": "# React Best Practices\n\nUse React Server Components by default..."
}

call_tool

Execute an MCP tool on a remote server. Use for results with type: "mcp_tool". For skills, use get_skill instead.

Parameters

Name	Type	Default	Description
endpoint	string	required	MCP server endpoint URL
tool	string	required	Tool name to call
args	string	required	JSON string of arguments
maxResponseChars	number	10000	Max response size, 1k–50k chars

Example

call_tool({
  endpoint: "https://send-email-mcp.example.com/mcp",
  tool: "send_email",
  args: '{"sender_email":"you@example.com","to_emails":["recipient@example.com"],"subject":"Hello"}'
})

Workflow

The typical agent workflow — whether using MCP or the REST API:

Call search_tools or GET /api/search with a natural language query.

Check the type

Each result has type: "mcp_tool" or "skill".

If skill

Call get_skill or GET /api/skill/:name to load instructions, then follow them.

If MCP tool

Call call_tool or POST /api/call with the endpoint, tool, and args from the search result.