Jira GitHub Manager CLI

CLI tool for managing Jira and GitHub integration with AI assistance via Gemini.

Prerequisites

• llm CLI tool installed (pip install llm or brew install llm)
• Jira API token (stored in .env as JIRA_API_TOKEN)
• GitHub personal access token (stored in .env as GH_SNIPPETS_TOKEN)
• Google Gemini API key (stored in .env as GEMINI_KEY)

Setup

1. Environment Variables

Create a .env file with:

JIRA_API_TOKEN=your_jira_token
GH_SNIPPETS_TOKEN=your_github_token
GEMINI_KEY=your_gemini_api_key

Get your keys:

• Jira: https://id.atlassian.com/manage-profile/security/api-tokens
• GitHub: https://github.com/settings/tokens
• Gemini: https://ai.google.dev/

config.yml

jira:
  server:
    address: https://company.atlassian.net
    type: cloud
  auth:
    type: basic
    user: [email protected]
  defaults:
    project: XXXX
    board: ""
    orgId: 45a1b0f9-50a8-4def-8936-c77c86e7cfac

github:
  auth:
    type: pat
  defaults:
    owner: "username"
    repo: "repo_name"

ai:
  provider: gemini
  model: gemini-2.0-flash
  options:
    codeExecution: false
    temperature: 0.1
g
automation:
  createPrOnProgress: false
  autoLinkIssues: true
  branchTemplate: "{project}-{issue-number}-{summary}"

2. Install LLM and Gemini Plugin

# llm is already installed if you have it globally
# Install Gemini plugin
llm install llm-gemini

3. Configure Gemini

# One-time setup
./setup-gemini.sh

This script:

• Reads your GEMINI_KEY from .env
• Stores it in LLM's secure configuration
• Tests the connection to Gemini API

Usage

Test Gemini Connection

llm -m gemini-2.0-flash "Say hello"

Interactive Chat with Gemini

llm chat -m gemini-2.0-flash

Use Different Gemini Models

# Fastest, cheapest
llm -m gemini-1.5-flash-8b-latest "Your prompt"

# Balanced (default)
llm -m gemini-2.0-flash "Your prompt"

# Most capable
llm -m gemini-2.5-pro-preview-05-06 "Your prompt"

Analyze Images

llm -m gemini-2.0-flash 'extract text' -a screenshot.png
llm -m gemini-2.0-flash 'describe' -a https://example.com/image.jpg

Code Execution

llm -m gemini-2.0-flash -o code_execution 1 \
  'use python to analyze this: [data]'

With Claude Code Integration

# In Claude Code, use Context7 to fetch docs while leveraging Gemini
llm -m gemini-2.0-flash 'Based on this API docs: [context], write code to...'

Configuration Files

• .env - API credentials (not tracked in git)
• .llmrc - LLM configuration (model defaults)
• setup-gemini.sh - Setup script for Gemini

Documentation

• Context7 Setup - Using Context7 with Claude
• Gemini Setup - Complete Gemini guide
• llm-gemini - Official plugin docs
• LLM - Main LLM tool docs

Models Available

| Model | Speed | Cost | Use Case | |-------|-------|------|----------| | gemini-1.5-flash-8b-latest | ⚡⚡⚡ | $ | Quick, lightweight tasks | | gemini-2.0-flash | ⚡⚡ | $$ | General purpose (default) | | gemini-2.5-pro-preview | ⚡ | $$$ | Complex reasoning |

Troubleshooting

"Key not found" error

# Re-run setup
./setup-gemini.sh

# Or manually verify key is set
llm keys list

API errors

# Check if model is available
llm models list | grep gemini

# Test with explicit key
export LLM_GEMINI_KEY=$(grep GEMINI_KEY .env | cut -d= -f2)
llm -m gemini-2.0-flash "test"

Rate limiting

Use a cheaper model:

llm -m gemini-1.5-flash-8b-latest "Your prompt"

Next Steps

1. Implement Jira integration functions
2. Add GitHub integration
3. Create CLI commands using Click or Typer
4. Use Gemini for AI-powered features

Development Notes

• Never commit .env file
• Keep API credentials secure
• Test changes with llm CLI before building main app
• Use Context7 for documentation-aware AI features

License

MIT