piyushfgk/pitools

PiTools MCP - Useful Tools for Any Prompt

❌ Without PiTools

• ❌ No up-to-date web search in your AI agent
• ❌ Manual tab-switching for research
• ❌ No easy way to add more tools to your AI workflow

✅ With PiTools

PiTools MCP brings real-time web search, social media posting, and more tools directly into your AI agent or coding assistant. Just add use pitools to your prompt in Cursor or any MCP-compatible client:

Search the web:

What is the latest version of Next.js? use pitools duckduckgo_search

Post to LinkedIn:

Post "Excited to share my new project update! #innovation #coding" to LinkedIn. use pitools linkedin_post_text

• 1️⃣ Write your prompt naturally
• 2️⃣ Tell the LLM to use pitools and specify the tool (e.g., duckduckgo_search, linkedin_post_text)
• 3️⃣ Get up-to-date answers or see your actions performed

No tab-switching, no outdated info, no manual research or posting.

🛠️ Getting Started

Requirements

• Node.js >= v18.0.0
• Cursor, Windsurf, Claude Desktop, or any MCP Client
• For LinkedIn tools: LINKEDIN_ACCESS_TOKEN environment variable set with a valid LinkedIn access token.

Install in Cursor

Go to: Settings → Cursor Settings → MCP → Add new global MCP server

Paste the following configuration into your Cursor ~/.cursor/mcp.json file (replace the path if needed). Make sure to replace YOUR_LINKEDIN_ACCESS_TOKEN, YOUR_INSTAGRAM_ACCESS_TOKEN, and YOUR_INSTAGRAM_USER_ID_FOR_POSTING with your actual credentials.

{
  "mcpServers": {
    "pitools": {
      "command": "npx",
      "args": ["-y", "@psachan/pitools@latest"],
      "env": {
        "LINKEDIN_ACCESS_TOKEN": "YOUR_LINKEDIN_ACCESS_TOKEN",
        "INSTAGRAM_ACCESS_TOKEN": "YOUR_INSTAGRAM_ACCESS_TOKEN",
        "INSTAGRAM_USER_ID_FOR_POSTING": "YOUR_INSTAGRAM_USER_ID_FOR_POSTING"
      }
    }
  }
}

• You can also use bunx or pnpm dlx if you prefer.
• If you prefer to set the environment variables system-wide, you can omit the env block here, but ensure they are accessible to the npx command.

🧰 Available Tools

• duckduckgo_search: Performs a web search using DuckDuckGo and returns the top results.

  • query (string, required): The search query string.
  • options (object, optional):
    • safeSearch (optional, default: MODERATE): STRICT | MODERATE | OFF.
    • region (optional, default: us-en): Search region (e.g., 'us-en', 'uk-en').
    • maxResults (optional): Maximum number of results to return.
    • time (optional): Time range for results (DAY, WEEK, MONTH, YEAR).

• linkedin_post_text: Posts a text update to LinkedIn.

  • commentary (string, required): The main text content of the post.

• linkedin_post_article: Posts an article (URL with commentary and title) to LinkedIn.

  • commentary (string, required): Text to accompany the article.
  • url (string, required): The URL of the article to share.
  • title (string, required): The title of the article.
  • thumbnailFilePath (string, optional): Local path to an image to use as the article thumbnail.
  • thumbnailAltText (string, optional): Alt text for the custom thumbnail image.

• linkedin_post_image: Posts an image with commentary to LinkedIn from a local file path.

  • commentary (string, required): Text to accompany the image.
  • filePath (string, required): The local path to the image file.

• linkedin_post_video: Posts a video with commentary to LinkedIn from a local file path.

  • commentary (string, required): Text to accompany the video.
  • filePath (string, required): The local path to the video file.

• linkedin_post_poll: Posts a poll with commentary to LinkedIn.

  • commentary (string, optional): Text to accompany the poll.
  • pollQuestion (string, required): The question for the poll.
  • pollOptions (array of strings, required): 2 to 4 options for the poll.

• instagram_post_image: Posts an image to Instagram from a publicly accessible URL with an optional caption.

  • imageUrl (string, required): Publicly accessible URL of the image to post.
  • caption (string, optional): Text to accompany the image.

(More tools coming soon!)

🧑‍💻 Development

Clone the project and install dependencies:

npm install

Build:

npm run build

Local Configuration Example

Make sure to replace tokens and IDs with your actual credentials if testing features.

{
  "mcpServers": {
    "pitools-local": {
      "command": "npx",
      "args": ["tsx", "/path/to/folder/pitools/src/index.ts"],
      "env": {
        "LINKEDIN_ACCESS_TOKEN": "YOUR_LINKEDIN_ACCESS_TOKEN",
        "INSTAGRAM_ACCESS_TOKEN": "YOUR_INSTAGRAM_ACCESS_TOKEN",
        "INSTAGRAM_USER_ID_FOR_POSTING": "YOUR_INSTAGRAM_USER_ID_FOR_POSTING"
      }
    }
  }
}

🧪 Testing with MCP Inspector

npx -y @modelcontextprotocol/inspector npx @psachan/pitools@latest

License

ISC