Agent Orchestrator

This project is an LLM coding agent orchestrator and AI software engineering platform designed to manage fleets of agents that autonomously solve issues, handle pull requests, and fix CI failures. It functions as an agentic CI/CD automator and parallel workflow manager, coordinating the end-to-end development lifecycle from initial ticket tracking to final code merging.

The system is distinguished by its modular plugin framework and isolated worktree management, which allow multiple agents to work on separate coding tasks simultaneously without file system conflicts. It utilizes role-based model routing to assign different large language models to orchestration and execution tasks, balancing high-level reasoning with processing speed.

The platform covers a broad range of capabilities, including automated CI remediation and code review loops that route failure reports and reviewer comments back to agents for iterative fixes. It provides session and workspace management via a centralized dashboard, featuring bidirectional terminal streaming, state-based session persistence, and integrated issue tracking.

The orchestrator is built with a plugin-based architecture that supports swappable components for AI models, execution runtimes, and notification gateways, and it can be deployed within Docker environments.

Features

AI Agent Orchestration - Provides a unified interface for managing the lifecycle and roles of multiple AI agents across several repositories.
CI/CD Automation Servers - Functions as an orchestrator that detects CI failures and automatically triggers agents to implement fixes.
Event-Driven Agent Loops - Triggers agent re-activation based on events like CI failures or new pull request comments.
Agent Lifecycle Management - Manages the creation, tracking, and cleanup of AI agent worker sessions based on PR state.
Coding Agent Orchestrators - Coordinates multiple AI coding agents in parallel, utilizing isolated worktrees to prevent file system conflicts.
Worktree Isolation - Uses separate git worktrees for each agent session to prevent file system conflicts during parallel development.
Programmatic Agent Spawning - Programmatically spawns new agent sessions for specific issues or prompts with custom plugin configurations.
AI Review Loop Management - Tracks pull request status and CI results to prompt agents to fix tests or address reviewer changes.
AI Software Engineering - Automates the end-to-end development lifecycle from issue tracking and coding to review and merging.
Coding Agent Integrations - Integrates AI agents with issue trackers and version control to automate the workflow from ticket to merged pull request.
Multi-Repository Orchestration - Provides a unified interface to control and monitor AI agents working across several different code repositories.
Parallel Coding Sessions - Spawns multiple AI agents to resolve separate issues simultaneously using isolated worktrees.
Automated Pull Request Reviewers - Automatically scans pull requests for review comments and triggers AI agents to address the feedback.
Agent Session Parallelization - Executes multiple concurrent AI agent coding sessions in isolation to process separate issues simultaneously.
Git Workspace Isolations - Provides each agent session with its own git worktree to prevent conflicts during parallel development.
CI Failure Remediation - Automatically detects failed CI builds and re-activates AI agents to implement the necessary fixes.
Agentic Feedback Routing - Routes failure reports and review comments directly to agents to trigger autonomous bug fixes.
Review Comment Synchronization - Monitors pull request comments for requested changes and automatically prompts the agent to apply fixes.
Pull Request Management - Coordinates the end-to-end lifecycle of code changes from pull request creation through CI tracking and merging.
Plugin-Based Architectures - Employs a modular architecture allowing the swapping of LLM providers, runtimes, and issue trackers.
CI Build Remediation - Monitors CI pipelines and automatically triggers AI agents to fix build failures and regressions.
Workflow Management Dashboards - Provides a centralized dashboard to launch, monitor, and review the work of multiple agents across repositories.
Agent Plugin Frameworks - Provides a pluggable architecture for swapping LLM models, execution runtimes, and external tool integrations.
Agent Runtimes - Executes agent processes using diverse backends such as tmux or standard system processes.
Agent Prompt Templates - Allows the injection of project-specific instructions and conventions into agent prompts to enforce coding standards.
Batch Agent Spawning - Launches multiple agent sessions for a list of issues simultaneously while skipping duplicate or active sessions.
Dynamic Model Swapping - Configures which AI model or plugin performs specific coding tasks and defines their file system permissions.
Fix Verification Workflows - Automatically marks issues as verified or failed after performing staging checks and updating the issue record.
Role-Based Model Assignment - Assigns different models to orchestration and execution roles to balance reasoning depth with speed.
Bidirectional Agent Streams - Provides bidirectional WebSocket streaming between the orchestrator and agent shells for real-time interaction.
AI Agent Session Managers - Provides tools to list, attach to, kill, and restore agent sessions and their associated worktrees.
Component Configuration - Integrates models, runtimes, issue trackers, and notification services through a central modular configuration file.
Issue Tracking Integrations - Integrates the orchestration process with external issue trackers to align automated tasks with project tickets.
Terminal Session Management - Manages the lifecycle of agent terminal sessions for real-time interaction via local terminals or a dashboard.
Session Attachment Tools - Allows users to connect to running agent sessions through native terminal emulators or a web-based dashboard.
Bidirectional Terminal Interfaces - Implements a live bidirectional terminal interface via WebSockets to monitor agent output and send commands.
Workflow Automation Triggers - Implements triggers that initiate automated agent actions in response to system events like CI failures or review comments.
Pull Request Status Trackers - Provides a visualization of agent-created pull requests including CI status, diff sizes, and review decisions.
Plugin Extensibility - Integrates external plugins from packages or local paths to add custom trackers or notification handlers.
Notification Routing Services - Aggregates and dispatches alerts across multiple channels by mapping alert levels to specific notifier instances.
Agentic Session Persistence - Persists active process metadata and workspace states to disk to enable session recovery after reboots.
AI Agent Role Assignments - Assigns specific roles such as orchestrator or executor to different LLMs to optimize for reasoning and speed.
Agent Execution Environments - Provides the ability to swap between different compute layers like tmux, Docker, or Kubernetes for agent execution.
Per-Session Workspace Isolations - Ensures agent session isolation using tmux windows or child processes to prevent cross-task interference.
Agent Configuration Files - Implements configuration files to define agent roles, runtimes, and isolation strategies.
Multi-Project Task Coordination - Links agent activities to external ticket systems and team identifiers across multiple projects.
Notification Dispatchers - Routes execution alerts to various external services including desktop notifications, chat platforms, and webhooks.
Project Progress Tracking - Monitors software development progress by integrating with issue trackers and version control pull requests.
Real-Time Event Notifications - Sends real-time status updates across multiple transport backends based on event priority.
Internal Telemetry Monitors - Captures and exposes internal operational state snapshots as JSON to power a live activity dashboard.
Agent State Telemetry - Writes internal component snapshots to disk to power a live dashboard of agent activities and system health.
Session Activity Monitors - Provides an interface for tracking and analyzing active agent development sessions, including branch and CI status.
AI Agent Activity Monitors - Provides real-time tracking of AI agent tools and sub-agent task progress via a live terminal view.
Orchestrator Session Management - Allows selecting between existing orchestrator instances or spawning new sessions to coordinate agent fleets.
Session Termination - Provides commands for shutting down running orchestrator agents and dashboards with options to purge session data.
Development Session Orchestrators - Provisions isolated workspaces and launches the orchestrator and web dashboard via local paths or repository URLs.

Dimillian/CodexMonitor

2,570View on GitHub

CodexMonitor is an AI agent orchestration interface designed for monitoring agentic workflows and managing remote daemon connections. It provides a web-based dashboard for coordinating AI agents across local workspaces and managing the execution of large language model tasks. The system distinguishes itself by integrating AI agents directly into git-based development workflows, synchronizing GitHub issues and pull requests with conversation threads. It uses branch worktree isolation to run tasks in separate physical directory copies, preventing state leakage between concurrent agent activitie

dagger/container-use

3,556View on GitHub

container-use is a containerized AI execution environment and code sandbox designed to provide a secure space for AI coding agents to execute commands and build applications. It functions as a workspace orchestrator that provisions isolated containers mapped to git branches, allowing multiple agents to operate in parallel without state conflicts or affecting the host system. The project serves as a Model Context Protocol server, bridging AI agents to containerized environments for standardized tool access. It enables a workflow for reviewing and merging changes made by agents within these iso

microsoft/vscode-copilot-chat

9,493View on GitHub

This project is an AI-powered IDE extension and LLM coding assistant that provides a conversational interface for generating, refactoring, and debugging code. It functions as an AI agent framework and a Model Context Protocol client, connecting AI models to external data sources and tools to automate complex development tasks. The system is distinguished by its use of autonomous AI agents capable of multi-step task execution, including the ability to read files, modify code, and run terminal commands iteratively. It supports recursive agent orchestration through subagent delegation and employ

livekit/livekit

19,358View on GitHub

LiveKit is a comprehensive framework for building and orchestrating real-time, multimodal AI agents that interact with users through voice, video, and text. It provides a centralized, event-driven architecture to manage the entire lifecycle of automated participants, from initialization and session state management to graceful shutdown. By utilizing a selective forwarding unit, the platform efficiently routes media streams between participants and agents, ensuring low-latency communication and secure, token-based authentication for all connections. The platform distinguishes itself through it

AgentWrapperagent-orchestrator

Features