SpecForge
SpecForge is a Model Context Protocol (MCP) server that connects an LLM client to a fully autonomous code-fixing pipeline. Point it at a GitHub issue, and it reads the repo, plans a fix, writes the code, tests it, and opens a pull request.
README
SpecForge
Turn GitHub issues into merged pull requests — autonomously.
SpecForge is a Model Context Protocol (MCP) server that connects an LLM client (like Claude Desktop) to a fully autonomous code-fixing pipeline. Point it at a GitHub issue, and it reads the repo, plans a fix, writes the code, tests it, and opens a pull request — without a human touching the keyboard in between.
Built for the CyOps Arena x MiniMax M3 Hackathon 2025.
How It Works
SpecForge orchestrates four stages, each handled by a specialized agent:
| Stage | Agent | What Happens |
|---|---|---|
| 1. Recon | Recon Agent | Reads repo structure, detects language, package manager, and test runner |
| 2. Plan | Planning Agent | Generates a file-level implementation plan from the issue description |
| 3. Implement | MiniMax M3 | Writes the actual code changes |
| 4. Review | Review Agent | Runs tests, self-critiques, and loops until the suite passes |
| 5. Ship | GitHub Agent | Creates a branch, commits changes, and opens a pull request |
You give it an issue URL. It gives you back a PR.
Prerequisites
| Requirement | Notes |
|---|---|
| Node.js | v18 or higher |
| npm | v9 or higher |
| CyOps account | Free signup at ai.cysic.xyz |
| GitHub Personal Access Token | Needs repo and pull_request scopes |
Installation
1. Install dependencies
cd mcp-server
npm install
npm run build
2. Configure environment variables
cp .env.example .env
Edit .env with your credentials:
# GitHub — token needs repo, pull_requests, and contents (read + write) scopes
GITHUB_TOKEN=ghp_xxxxxxxxxxxxxxxxxxxx
# CyOps Gateway URL (from your ai.cysic.xyz workspace)
CYOPS_GATEWAY_URL=https://your-instance.cysic.xyz
# CyOps API key (Settings → API Keys, inside your workspace)
CYOPS_API_KEY=cyops_xxxxxxxxxxxxxxxxxxxx
3. Set MiniMax M3 as your model in CyOps
- Open your CyOps workspace and click Open Gateway
- Go to Settings → LLM Proxy
- Set each agent's model to MiniMax M3
This unlocks the hackathon's 80% discount on token pricing.
4. Connect to Claude Desktop
Open your Claude Desktop config file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
Add SpecForge as an MCP server:
{
"mcpServers": {
"specforge": {
"command": "node",
"args": ["/absolute/path/to/mcp-server/dist/index.js"]
}
}
}
Restart Claude Desktop. SpecForge should now appear in your available tools.
Usage
In Claude Desktop, just describe what you want fixed:
Use specforge to fix https://github.com/org/repo/issues/42
SpecForge runs the full pipeline and returns a pull request URL when finished. No further input required unless you want to check progress mid-run.
Available Tools
specforge_run
Kicks off the full pipeline for a given issue.
{
issue_url: string; // Required — full GitHub issue URL
strategy_profile?: string; // Optional — CyOps strategy (default: "Default RLCR (Claude → MiniMax M3)")
max_wait_minutes?: number; // Optional — timeout in minutes (default: 10)
}
specforge_status
Checks the status of an in-progress run.
{
run_id: string; // CyOps run ID, returned from specforge_run
project_id: string; // CyOps project ID, returned from specforge_run
}
Architecture
specforge_run (MCP tool)
│
├── parseIssueUrl() → Parse the GitHub issue URL
├── fetchIssue() → Read issue body + comments
│
├── runRecon() → Recon Agent
│ ├── fetchRepoContext()
│ ├── readFile() × N → package.json, tsconfig, etc.
│ └── detect test runner, package manager
│
├── createProject() → CyOps: create new project
├── generatePlan() → CyOps: generate plan from requirement
├── startRun() → CyOps: begin Implement → Review loop
├── pollRun() → CyOps: wait for completion
│
├── getArtifacts() → List changed files
├── readWorkspaceFile() × N → Read each changed file
├── createBranch() → GitHub: create new branch
├── commitFileToGitHub() × N → GitHub: push each changed file
└── openPullRequest() → GitHub: open pull request
Project Structure
mcp-server/
├── src/
│ ├── index.ts # MCP server entry point + tool handlers
│ ├── github.ts # GitHub API client
│ ├── cyops.ts # CyOps Gateway API client
│ └── agents/
│ ├── recon.ts # Recon Agent — repo understanding
│ └── orchestrator.ts # Main pipeline coordinator
├── package.json
├── tsconfig.json
└── README.md
Hackathon
CyOps Arena x MiniMax M3 — June 2025
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.