Skip to content
MCP server

Widal.Code

Shows what your AI coding agents spend on Claude Code, Codex and Cowork, and where to cut

Runs locally
Install
Pricing
Freemium
Platforms
macOS, CLI, vscode
Runs
Locally
Version
0.3.0
Listing updated
today

Similar MCP servers

About Widal.Code

Widal.Code is a freemium MCP server for developer tools. It runs locally on your machine and is started by your AI client.

The README says Widal.Code reads the session logs that Claude Code, Codex and Cowork keep on your machine and turns them into token counts, costs at list prices and plan-limit usage. Nothing leaves the computer unless you link it to a team board, which joins the spend with the team's GitHub pull requests.

In Claude Code it adds a status line showing plan limits, context size and prompt cache timing, and it proposes cheaper model or effort settings. Each proposal can be applied with one command and undone with another. A separate set of checks reviews agent settings, actions and changed code for security risks, and can run read-only checks on AWS, Vercel and GitHub when asked.

The tool suits developers who use AI coding agents daily, and engineering leads who want to see team agent costs. The team board is a paid add-on at $5 per active person per month. The local tool is free and MIT-licensed, and so far it runs on macOS only according to the README.

Install

Run in a terminal

claude mcp add widal-code -- npx -y https://github.com/Widal-Inc/widal-code/releases/download/v0.3.0/widal-code.mcpb

Add to claude_desktop_config.json

{
  "mcpServers": {
    "widal-code": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/Widal-Inc/widal-code/releases/download/v0.3.0/widal-code.mcpb"
      ]
    }
  }
}
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "widal-code": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/Widal-Inc/widal-code/releases/download/v0.3.0/widal-code.mcpb"
      ]
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "widal-code": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/Widal-Inc/widal-code/releases/download/v0.3.0/widal-code.mcpb"
      ]
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "widal-code": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/Widal-Inc/widal-code/releases/download/v0.3.0/widal-code.mcpb"
      ]
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "widal-code": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/Widal-Inc/widal-code/releases/download/v0.3.0/widal-code.mcpb"
      ]
    }
  }
}

Using several servers? Build one config file with the MCP config generator.

What it does

Session and cost reports
Breaks token use and list-price cost down by call, session, ticket, project, model and account.
Claude Code status line
Shows plan limits, context size, prompt cache timing and open security risks at the bottom of each session.
Cache and context waste alerts
Flags lost prompt caches, oversized context and models or effort levels that cost more than the work needs.
Reversible setting changes
Proposes config changes that are applied with one command and undone with another.
Agent security checks
Reviews agent settings, logged actions and changed code, mapping findings to chosen compliance frameworks.
Team board
Optional paid board that adds team spend, cost per merged pull request and security findings across linked computers.

Good for

  • Finding which sessions, subagents or models drive most of a developer's agent spend
  • Checking how much of a Claude Code plan's limits is left before reset
  • Reviewing agent settings for risky permissions, secret file reads or leaked keys
  • Tracking the cost of each merged pull request across an engineering team
  • Answering spend questions in the Claude desktop app through the MCP extension

Pricing

Freemium, from $5 a month

Free on one computer; team board $5 per active person per month

Prices change. Check code.widal.com before you buy.

Widal.Code FAQ

Does Widal.Code send my prompts or files anywhere?

The README says it sends nothing on its own. When a computer is linked to the team board, it sends call metadata, folder and branch names, and security finding descriptions, but never the text of prompts, answers, tool results or files.

How much does it cost?

The single-person tool is free and open source under the MIT licence. The team board costs $5 a month for each active person, with no free trial.

Which platforms does it run on?

The README says it has so far run on macOS only, requires Node 22 or newer, and is also available as a Claude desktop extension.

Do the dollar figures show what I actually pay?

No. The README states the dollar values are list-price API equivalents. On a subscription such as Claude Pro or Max, the plan's limit share is what counts instead.

Is Widal.Code free?

Widal.Code has a free plan, with paid plans for more usage or features. Paid plans start at $5 a month.

How do I add Widal.Code to Claude Code?

Run this command in your terminal: claude mcp add widal-code -- npx -y https://github.com/Widal-Inc/widal-code/releases/download/v0.3.0/widal-code.mcpb

What are the best alternatives to Widal.Code?

The closest alternatives to Widal.Code are Kestrel, Actvt, AI Usage, CodeBurn and context-doctor.

Where this comes from

  • Description: written from the project's README, the project's website and its MCP Registry entry.
  • Install steps: from the official MCP Registry entry.

How AIPoweredBot checks servers