AgentGate — Supervised Access for AI Agents to Jira
AgentGate lets AI agents read, create, and update Jira issues through a safe, human-approved workflow. Every write operation goes through an approval queue — nothing touches Jira until a human reviews and approves it.
Documentation Sections
Overview
Learn how AgentGate works, its architecture, key features, and how it compares to alternatives.
Getting Started
Install the Forge app, generate a token, and connect your first AI agent in 10 minutes.
MCP Server
Connect to Claude Desktop, Claude Code, Cursor, and other MCP clients.
CLI
Install and use the jd command-line tool for terminal-based workflows.
Claude Code Plugin
Auto-context, /jira skill, and session continuity for Claude Code.
Approval Workflow
How the human-in-the-loop review process works.
Tools Reference
Complete reference for all 15+ AgentGate tools with parameters and examples.
Troubleshooting
Common issues and how to fix them.
Quick Links
- Atlassian Marketplace — Install the Forge app
- npm: @orbiscend/jd-mcp — MCP server package
- npm: @orbiscend/jd-cli — CLI package
- Docker: gustavorbiscend/jd-mcp — Docker image
- GitHub: agentgate-claude-plugin — Claude Code plugin