# GitHub Issue MCP [Health: Active]

**Category:** 🔄 Version Control  
**Repository:** https://github.com/sarim-aliii/github-issue-mcp  
**GitHub Stars:** 0  
**npm Downloads (last month):** 171  
**Views:** 2  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/github-issue-mcp

## Description
AI-powered MCP server for generating GitHub issues and detecting semantic duplicates.

## Tools
Capabilities this server exposes over MCP:

- **generate_issue** — Generate a structured GitHub issue using the issue description, issue details, and repository labels.
- **create_github_issue** — Create a new issue in a GitHub repository
- **list_github_issues** — List open issues in a GitHub repository
- **list_github_labels** — List labels available in a GitHub repository
- **check_duplicate_issue** — Find the most likely existing open GitHub issues that could be related to a proposed issue. This tool retrieves open issues and performs cheap lexical candidate ranking. Treat the returned issues as candidates only; semantic duplicate determination must be done by the AI. Do not call list_github_issues separately for duplicate checking.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "github-issue-mcp": {
    "command": "npx",
    "args": ["-y","github-issue-mcp"],
    "env": {
      "GITHUB_TOKEN": "",
      "GITHUB_OWNER": "",
      "GITHUB_REPO": "",
      "GEMINI_API_KEY": ""
    }
  }
}
```

**Requires environment variables:** `GITHUB_TOKEN`, `GITHUB_OWNER`, `GITHUB_REPO`, `GEMINI_API_KEY` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What the GitHub Issue MCP MCP server does

GitHub Issue MCP MCP server supports a controlled workflow for managing issues in one configured GitHub repository. It can turn supplied bug information into a structured issue draft, retrieve open issues and repository labels, identify likely duplicate candidates, and create a new issue after approval.

The generation step accepts details such as a description, reproduction steps, expected and actual behavior, environment information, and additional context. It returns a proposed title, formatted body, and labels. The server also supports label retrieval so an application can inspect the labels available in the target repository.

## How it works

The GitHub Issue MCP MCP server runs as an MCP process over standard input and output. GitHub operations use the GitHub REST API, while Google Gemini handles issue generation and the final semantic review of duplicate candidates.

Duplicate checking uses two stages. First, `check_duplicate_issue` retrieves open issues and applies inexpensive lexical ranking to produce a smaller candidate set. Those results are only possible matches, not proof of duplication. Gemini then performs the semantic comparison. Do not call `list_github_issues` separately for duplicate checking; use the dedicated duplicate tool.

The intended workflow keeps persistent writes outside autonomous model access. After duplicate analysis, an application can collect more details, generate a draft, show it to the user, and request approval. Only then should the client invoke `create_github_issue`. The tool is deliberately not available to the autonomous Gemini agent.

## Setup and configuration

Install the published npm package with `npm install github-issue-mcp`, or launch the MCP CLI directly with `npx -y github-issue-mcp`. Configure an MCP-compatible client to start that command and communicate over stdio.

The server requires `GITHUB_TOKEN`, `GITHUB_OWNER`, `GITHUB_REPO`, and `GEMINI_API_KEY`. `GITHUB_OWNER` and `GITHUB_REPO` identify the repository used by the GitHub tools. Optional settings include `GEMINI_MODEL`, `GEMINI_MAX_RETRIES`, `GEMINI_RETRY_BASE_DELAY`, and `DEBUG`; the documented defaults are `gemini-3.6-flash`, `3`, `1000` milliseconds, and `false` respectively.

Keep GitHub tokens and Gemini API keys out of source control and do not commit the `.env` file.

## Tools and capabilities

- `generate_issue`: Builds a structured issue from supplied issue information.
- `check_duplicate_issue`: Ranks likely matches among existing open issues for later semantic review.
- `list_github_issues`: Retrieves open issues and excludes pull requests.
- `list_github_labels`: Retrieves labels from the configured repository.
- `create_github_issue`: Creates the approved issue in GitHub.

The GitHub Issue MCP MCP server also includes retry handling for temporary Gemini failures and detects daily Gemini quota conditions without unnecessary retries. Model selection and retry behavior are configurable.

## Limitations and notes

Duplicate scores and ranked results are candidates only. The application or AI must make the semantic duplicate decision; lexical ranking alone is not sufficient. The repository configuration is supplied through environment variables, so the material describes one configured owner and repository rather than a multi-repository selection interface.

Creating an issue is an external write operation. The documented design requires explicit human approval before the client calls the creation tool, and the Gemini agent cannot invoke that write operation autonomously.

_Full upstream README: https://allmcps.com/mcp/github-issue-mcp/readme_