DevAssist - AI-Powered Developer Assistant

A Python CLI application that aggregates context from multiple developer tools (GitHub, Jira, Slack) and uses AI to provide:

• Morning Briefs - Consolidated summaries of your PRs, issues, and messages
• Interactive Chat - Ask questions about your work in natural language
• Background Daemon - Scheduled briefs at 8am, 1pm, and 5pm

Features

• GitHub Integration - PRs needing review, issues assigned to you, notifications
• Jira/Atlassian Integration - Open issues, sprint status, recent updates
• Slack Integration - Unread messages, mentions, channel activity
• Interactive REPL - devassist chat for continuous conversation
• Background Daemon - Runs in background, generates scheduled briefs

Quick Start

1. Clone and Install

# Clone the repository
git clone https://github.com/ayush17/notebooks-ai-agent.git
cd notebooks-ai-agent

# Create virtual environment
python -m venv .venv
source .venv/bin/activate  # Linux/macOS
# or: .venv\Scripts\activate  # Windows

# Install in development mode
pip install -e ".[dev]"

2. Install MCP Servers

# GitHub MCP (required)
npm install -g @modelcontextprotocol/server-github

# Atlassian MCP: `npx -y mcp-remote https://mcp.atlassian.com/v1/mcp` (no global install required)
# Same transport as Cursor `mcp.json`; Node.js 18+ on PATH is enough.

3. Configure Credentials

Option A: Interactive Setup

devassist setup init

Option B: Manual Configuration

Create ~/.devassist/.env:

mkdir -p ~/.devassist

cat > ~/.devassist/.env << 'EOF'
# Claude AI (via Anthropic API)
export ANTHROPIC_API_KEY="your-anthropic-key"

# OR Claude on Vertex AI (Red Hat)
export CLAUDE_CODE_USE_VERTEX=1
export CLOUD_ML_REGION=us-east5
export ANTHROPIC_VERTEX_PROJECT_ID=your-gcp-project

# GitHub (required)
# Get token: https://github.com/settings/tokens
# Scopes needed: repo, notifications, read:user
export GITHUB_PERSONAL_ACCESS_TOKEN="ghp_xxx"

# Atlassian API (optional — only for `devassist brief` Jira/Confluence adapters;
# `devassist ask` / `chat -s atlassian` uses remote MCP and does not need these)
# Get token: https://id.atlassian.com/manage-profile/security/api-tokens
export ATLASSIAN_BASE_URL="https://your-site.atlassian.net"
export ATLASSIAN_EMAIL="your-email@example.com"
export ATLASSIAN_API_TOKEN="your-atlassian-token"
EOF

chmod 600 ~/.devassist/.env

4. Test It

# Load environment
source ~/.devassist/.env

# One-off question
devassist ask "What PRs need my review?" -s github

# Interactive chat
devassist chat -s github,atlassian

# Check status
devassist setup status

Usage

Ask Command (One-off Questions)

# GitHub queries
devassist ask "What PRs need my review?" -s github
devassist ask "Search for PRs where I'm a reviewer using is:pr is:open review-requested:@me" -s github

# Jira queries
devassist ask "What are my open Jira issues?" -s atlassian

# Combined
devassist ask "Give me a morning brief" -s github,atlassian

Chat Command (Interactive REPL)

devassist chat -s github,atlassian

Available commands in chat:

• /help - Show help
• /servers - List connected MCP servers
• /tools - List available tools
• /clear - Clear conversation history
• /quit - Exit

Background Daemon

# Start in foreground (for testing)
./scripts/start_daemon.sh

# Start in background
./scripts/start_daemon.sh -b

# Stop daemon
./scripts/stop_daemon.sh

# View logs
tail -f ~/.devassist/daemon.log

# View latest brief
cat ~/.devassist/briefs/latest.md

The daemon generates briefs at:

• 8:00 AM
• 1:00 PM
• 5:00 PM

Architecture

src/devassist/
├── cli/           # Typer CLI commands (ask, chat, setup)
├── core/          # Business logic (aggregator, ranker, brief_generator)
├── adapters/      # Context source adapters (gmail, slack, jira, github)
├── mcp/           # MCP client and server registry
├── orchestrator/  # LLM orchestration agent
├── ai/            # Vertex AI integration
├── preferences/   # Preference learning (planned)
└── models/        # Pydantic data models

MCP Servers

Server	Package	Purpose
GitHub	@modelcontextprotocol/server-github	PRs, issues, repos
Atlassian	mcp-remote → https://mcp.atlassian.com/v1/mcp	Jira, Confluence (hosted MCP)

Troubleshooting

"No MCP servers configured"

• Run devassist setup status to check configuration
• Ensure environment variables are set: source ~/.devassist/.env

Atlassian MCP slow or auth issues

• First run downloads mcp-remote; ensure outbound HTTPS is allowed
• Authentication follows Atlassian’s remote MCP flow (browser/OAuth as prompted by the connector)
• Large Jira searches may take 30–60 seconds

GitHub MCP asks for repo details

• Use specific search syntax: "Search for PRs using is:pr is:open review-requested:@me"
• The LLM needs guidance on GitHub search queries

Command not found: devassist

• Ensure you've activated the venv: source .venv/bin/activate
• Reinstall: pip install -e .

Development

# Run tests
pytest

# Run with coverage
pytest --cov=devassist

# Type checking
mypy src/

# Linting
ruff check src/

Requirements

• Python 3.11+
• Node.js 18+ (for MCP servers)
• Claude AI access (one of the following):
  • Anthropic API key from console.anthropic.com (paid)
  • OR Red Hat employees: Use Vertex AI with ANTHROPIC_VERTEX_PROJECT_ID=itpc-gcp-ai-eng-claude
• GitHub: Personal Access Token (free) - Create here
• Jira/Atlassian: Access to an Atlassian Cloud site (OAuth - browser login on first use)

Environment Variables Reference

Variable	Required	Description
ANTHROPIC_API_KEY	Yes*	Anthropic API key for Claude
CLAUDE_CODE_USE_VERTEX	Yes*	Set to 1 for Vertex AI
ANTHROPIC_VERTEX_PROJECT_ID	Yes*	GCP project ID for Vertex AI
GITHUB_PERSONAL_ACCESS_TOKEN	Yes	GitHub PAT with repo, notifications scopes
ATLASSIAN_BASE_URL	No	Optional; for devassist brief Jira adapter (not MCP remote)
ATLASSIAN_EMAIL	No	Optional; same
ATLASSIAN_API_TOKEN	No	Optional; same
SLACK_BOT_TOKEN	No	Slack bot token (xoxb-...)
SLACK_TEAM_ID	No	Slack workspace ID

*Either Anthropic API key OR Vertex AI config required

License

MIT License - see LICENSE file for details.

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Run tests: pytest
5. Submit a pull request