Jira Agentic Assistant

AI-powered Jira issue triage and weekly report generation. Configure everything through a local web UI — Jira credentials, LLM provider, your role, and write-back permissions.

Features

• Automated Triage: Categorize, summarize, find similar issues, and propose solutions using AI
• Weekly Reports: Generate markdown reports with trends, metrics, and recommendations
• Role-Based Analysis: 6 roles (Product Owner, Developer, PM, QA, Tech Lead, Scrum Master) that customize AI perspective
• Multiple LLM Providers: OpenAI, Ollama (local), or any OpenAI-compatible API
• Scheduling: Automatic triage and report generation via cron
• Write Mode: Optionally post triage comments back to Jira
• Local-First: All data stored locally in JSON files

Tech Stack

• Language: TypeScript 5.7+
• Monorepo: pnpm workspaces
• AI: LangGraph.js + LangChain.js
• Jira: jira.js v5
• Web UI: Next.js 15 + shadcn/ui + Tailwind CSS 4
• Storage: node-json-db
• Scheduling: node-cron

Quick Start

Prerequisites

• Node.js >= 20
• pnpm >= 9

Installation

# Clone the repository
git clone <repo-url>
cd jira-agentic-assistant

# Install dependencies
pnpm install

# Build the core package
pnpm --filter @jira-assistant/core build

# Start the development server
pnpm dev

Open http://localhost:3000 and complete the setup wizard.

Setup Wizard

1. Jira Connection: Enter your Jira Cloud URL, email, API token, and project key
2. LLM Provider: Choose OpenAI, Ollama, or OpenAI-compatible and configure credentials
3. Role Selection: Pick your role to customize AI analysis
4. Write Mode: Choose read-only or write-to-Jira

See docs/jira-pat-guide.md for Jira API token setup.

Project Structure

jira-agentic-assistant/
├── packages/core/          # @jira-assistant/core - Business logic
│   ├── src/
│   │   ├── agents/         # LangGraph triage & report agents
│   │   ├── config/         # Zod schemas & defaults
│   │   ├── db/             # node-json-db repositories
│   │   ├── jira/           # Jira client, reader, writer
│   │   ├── llm/            # LLM factory & prompts
│   │   ├── reports/        # Markdown report generator
│   │   ├── scheduler/      # node-cron job manager
│   │   └── types/          # TypeScript type definitions
│   └── vitest.config.ts
├── apps/web/               # @jira-assistant/web - Next.js dashboard
│   └── src/
│       ├── app/            # Pages & API routes
│       ├── components/     # UI components
│       ├── hooks/          # React hooks
│       └── lib/            # Utilities
└── data/                   # Runtime data (gitignored)
    ├── db/                 # JSON database files
    └── reports/            # Generated markdown reports

Usage

Manual Triage

Click "Run Triage" on the dashboard to fetch recent issues from Jira and analyze them with AI. Each issue gets:

• Category: bug, feature-request, improvement, question, task
• Severity: critical, high, medium, low
• Effort Estimate: XS, S, M, L, XL
• AI Summary: Concise 2-3 sentence summary
• Similar Issues: Previously triaged issues with similarity scores
• Proposed Solution: Actionable recommendation

Manual Report

Click "Generate Report" to create a weekly markdown report covering:

• Key metrics and category breakdowns
• Severity analysis
• Critical items requiring attention
• Trends and recommendations

Reports are saved as markdown files in data/reports/.

Scheduling

Enable automatic triage and report generation in Settings > Scheduling:

• Triage: Default every 4 hours (0 */4 * * *)
• Reports: Default Monday 9am (0 9 * * 1)

Development

# Build core package
pnpm --filter @jira-assistant/core build

# Watch mode for core
pnpm --filter @jira-assistant/core dev

# Run Next.js dev server
pnpm dev

# Run tests
pnpm test

License

APACHE 2.0