Roadmap

• Core functionality with projects with tasks processed by agents
• Local AI integration
• Dockerize the application
• Agents to collect and preprocess data from Jira (summary from Jira issue comments, sensitive information removal)
• GitHub and GitLab pull request integration (pull diffs and perform code review analysis)
• GitHub ongoing feedback processing (take in account previously provided feedback after new commits)
• Slack notification integration
• GitHub probot integration for live updates
• LLM api calls cache
• Process GitHub updates without probot (by cronjob)
• GitLab ongoing feedback processing (take in account previously provided feedback after new commits)
• Build UI for projects, tasks, and agents management
• OS notification agent
• Propose pull request changes (from AI-created new branch into PR branch)
• A whole codebase RAG with Jira context based on commit messages
• Create task from Jira issue, and let AI open pull request with task completed

Actions executed automatically

• (every 5 min) Get new requests from GitHub and create tasks for them
• (every 5 min) Get new requests from GitLab and create tasks for them
• (every 5 min) Process all tasks that are ready to be processed

Interactive Commands (docker-compose exec php php bin/console agi:*)

# Create new project
agi:project:create

# Create new agent config
agi:agent:create

# Add configured agent to existing project
agi:project:agents:add

# Process tasks in ready to process state
agi:tasks:process

# Check for new requests from GitHub, and create tasks for each new request
agi:github:check-new-requests

# Create task for a specific pull request (by URL)
agi:github:create-pr-task