# codev [Health: Active]

**Category:** 👤 Customer Data Platforms  
**Repository:** https://github.com/flxbl-io/codev-mcp  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/codev

## Description
Connect AI agents to codev for Salesforce projects, builds, releases, validations, and insights.

## Claude Desktop Quick Installation
Remote MCP endpoint (confidence: high). Install path detected from listing signals. Add as a URL/SSE server in your client:

```json
"mcpServers": {
  "codev": {
    "url": "https://docs.flxbl.io/flxbl/sfp/api-reference/mcp"
  }
}
```

## Documentation

## What codev does

The codev MCP server gives AI agents access to live delivery information for Salesforce projects managed in codev. It provides project and repository configuration, build and release progress, deployment state by environment, and the status of release candidates. Agents can also inspect pull-request validations, test and deployment results, code analysis, compliance findings, and review status.

Delivery context can include work items from connected Jira, Azure DevOps, or GitHub systems. The service also exposes registered environments, Salesforce orgs, pool capacity, assignments, and review environments. Project-level delivery performance insights are available for investigation and reporting.

## How it works

Each codev tenant exposes a remote MCP endpoint at `https://<your-tenant>.flxbl.io/sfp/api/mcp`. MCP clients connect to that URL over Streamable HTTP. Interactive clients discover the tenant’s OAuth 2.1 authorization server and open a browser sign-in flow, so credentials should not be placed in client configuration.

Tool calls use the signed-in user’s codev role. Pool checkouts are associated with that user’s email. Controlled automation can instead send a codev application token in an `Authorization: Bearer` header. Application tokens operate independently of a specific interactive user and should be stored in a secret manager.

## Setup and configuration

Replace `<your-tenant>` with the tenant subdomain in the MCP endpoint. Claude Code can be configured with `claude mcp add --transport http codev <URL>`, followed by sign-in through `/mcp`. Codex uses `codex mcp add codev --url <URL>` and `codex mcp login codev`.

Cursor users can install codev from the Cursor Marketplace and provide the tenant subdomain when prompted. Other HTTP-capable clients can use an MCP configuration containing the endpoint URL and no embedded credentials. In VS Code with GitHub Copilot, that configuration belongs in `.vscode/mcp.json`; Claude Desktop and claude.ai can add the URL as a custom connector.

## Tools and capabilities

The codev MCP server supports agents that need to:

- Discover Salesforce projects and repository settings.
- Track builds, release candidates, releases, and environment deployment state.
- Review pull-request validation, deployment, test, analysis, compliance, and review results.
- Read linked Jira, Azure DevOps, and GitHub work items.
- Examine Salesforce orgs, environments, pool capacity, assignments, and review environments.
- Query project delivery performance insights.
- Fetch, extend, or release a pool environment for the signed-in user.

## Limitations and notes

Most available operations do not change state. Pool environment fetch, extension, and release are the stated state-changing actions, and interactive users can manage only their own assignments. The remote MCP implementation is delivered with the codev product; this repository contains registry metadata and connection documentation rather than a duplicated implementation. Access remains subject to the permissions granted by the tenant and authenticated identity.

_Full upstream README: https://allmcps.com/mcp/codev/readme_