# brief bar_

> brief bar_ is an open-source MCP server for procurement. MCP server that scores digital product briefs against an open ruleset, with your own model. It runs as a hosted remote server, so there is nothing to install locally.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** API, CLI, Web, self-hosted
- **Website:** https://briefs.welance.com/integrate.html
- **Repository:** https://github.com/welance/perfect-brief
- **MCP Registry name:** com.welance/perfect-brief
- **Categories:** [procurement](https://ai-powered.com/categories/procurement)
- **Listing updated:** 2026-11-18

## About

The README describes a scoring service for digital product briefs that separates two outputs: a numeric score for how well a brief is written, and a publish gate that decides whether the brief may go live. Scoring is done by deterministic code against a versioned set of fourteen weighted rules, and every verdict carries a verbatim quote from the brief as evidence.

Through the hosted MCP endpoint, the assistant's own model does the judging using the service's judge prompt, and the server validates those verdicts and performs the aggregation. The site says no model of the vendor runs in this route, and that results are advisory: the Welance Directory scores a brief again before publishing it.

The bar is aimed at clients and suppliers who want expectations written down before committing to work together. The site is explicit that it measures articulation rather than whether an idea is sound.

## Features

- **Gate and score split:** A numeric score is reported alongside a separate publish gate that can refuse a brief on policy grounds.
- **Evidence-backed verdicts:** Each rule verdict includes a verbatim quote from the brief, so results can be inspected later.
- **Client-side judging:** The connected assistant's model performs the judgement while the server handles validation and arithmetic.
- **Versioned ruleset:** Results include the ruleset version, so a score can be tied to the exact rules that produced it.
- **Deterministic checks:** A check_brief tool runs contact, budget, timeline and placeholder checks without any model involved.
- **Open source core:** The rules and engine are published under the MIT licence in the project repository.

## Good for

- Checking a digital product brief for missing budget, timeline or problem definition before sending it to suppliers
- Getting a gate-first list of fixes to make a brief publishable
- Writing a brief from a rules-based template in an AI assistant
- Auditing which ruleset version judged a brief for later review

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http perfect-brief https://briefs.welance.com/mcp
```

### Claude Desktop (Settings → Connectors → Add custom connector)

```sh
https://briefs.welance.com/mcp
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "perfect-brief": {
      "url": "https://briefs.welance.com/mcp"
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "perfect-brief": {
      "type": "http",
      "url": "https://briefs.welance.com/mcp"
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "perfect-brief": {
      "serverUrl": "https://briefs.welance.com/mcp"
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "perfect-brief": {
      "type": "http",
      "url": "https://briefs.welance.com/mcp"
    }
  }
}
```

## Tools

- `get_rules`: Returns the live catalogue of criteria, weights, gate, bands and version.
- `get_brief_template`: Returns a Markdown brief built from the live rules.
- `get_judge_prompt`: Returns the judge prompt and verdict JSON Schema for your model.
- `get_improvement_prompt`: Returns a gate-first working order to improve the brief.
- `check_brief`: Runs deterministic checks only; semantic coverage is incomplete.
- `aggregate_verdicts`: Validates your model's verdicts and aggregates them.
- `bar_status`: Returns the live ruleset version and capabilities.

## FAQ

### Is the hosted MCP server free to use?

The Welance site and README describe the MCP server at briefs.welance.com/mcp as free, with no model of the vendor running for it.

### Does the MCP server give an authoritative decision?

No. The README says aggregated results are advisory and cannot authorise publication; the Welance Directory scores each brief itself.

### Which assistants can connect to it?

The README lists setup steps for Claude, Claude Code, ChatGPT, Cursor and VS Code, plus a local stdio option for other MCP clients.

### Can I use my own OpenRouter key with the local server?

The site says setting PB_LLM_KEY to your own OpenRouter key makes the judge run on your account.

### Is brief bar_ free?

Yes. brief bar_ is open source, so you can use and self-host it for free.

### How do I add brief bar_ to Claude Code?

Run this command in your terminal: claude mcp add --transport http perfect-brief https://briefs.welance.com/mcp

### What are the best alternatives to brief bar_?

The closest alternatives to brief bar_ are MCP CPG Human in the Loop (HITL), Eclipse Payseal, Eclipse Procure MCP and Florida Bid Intel MCP.

## 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.

## Alternatives

- [MCP CPG Human in the Loop (HITL)](https://ai-powered.com/mcp/mcp-cpg-human-in-the-loop-hitl): MCP CPG Human in the Loop (HITL) is an MCP server for procurement.
- [Eclipse Payseal](https://ai-powered.com/mcp/eclipse-payseal): Eclipse Payseal is an open-source MCP server for finance and payments.
- [Eclipse Procure MCP](https://ai-powered.com/mcp/eclipse-procure-mcp): Eclipse Procure MCP is an MCP server for procurement.
- [Florida Bid Intel MCP](https://ai-powered.com/mcp/florida-bid-intel-mcp): Florida Bid Intel MCP is an MCP server for legal and government.

---
Source: https://ai-powered.com/mcp/brief-bar
