Releases: fuzzylabs/sre-agent
0.2.1
What's Changed
🔄 Full Agent Rewrite with Pydantic AI
The core agent has been completely rewritten using Pydantic AI, replacing the previous implementation. This brings stronger type safety, cleaner tool definitions, and a more maintainable architecture.
🗑️ Removed Self-Hosted MCP Servers
The self-hosted Slack MCP server has been removed in favour of remote MCP servers. This significantly simplifies the deployment footprint — no more Node.js services to manage alongside the agent.
🧪 Evaluation Suite
A new evaluation suite has been added using Opik to measure agent quality:
- Tool extraction — verifies the agent selects the right tools for a given incident
- Root cause accuracy — measures how well the agent identifies the root cause
- Diagnosis quality — end-to-end assessment of the agent's diagnosis
🚀 ECS Deployment
Added support for deploying the agent on AWS ECS, with updated configuration and deployment docs.
🖥️ CLI & UX Improvements
Overhauled the CLI experience with better output formatting, a demo GIF, and updated documentation.
📖 Documentation
- Refreshed README with updated architecture and usage instructions
- New
DEVELOPMENT.mdguide - Added
RELEASE.mdand automated PyPI publishing via GitHub Actions
0.1.0
New Feature: Command-Line Interface
We're excited to introduce the SRE Agent CLI - a powerful command-line interface that brings AI-powered site reliability engineering directly to your terminal.
What's New
Interactive Shell
- Launch the agent with a simple
sre-agentcommand - Persistent command history for easy navigation
- Rich terminal UI with colour-coded output and formatted panels
Guided Setup
- First-time setup walks you through essential configuration
- Automatic AWS credentials and cluster connection testing
- Service discovery and selection from your Kubernetes cluster
- Validates all credentials before proceeding
Core Commands
diagnose [service]- Run AI-powered diagnostics on any service in your clusterconfig- Interactive configuration menu for all settingsstatus- View current connection and configuration statushelp- Access command documentation and examples
Configuration Management
- AWS Integration: Connect to EKS clusters with automatic kubectl configuration
- GitHub Access: Set up repository monitoring and issue creation
- Model Provider: Configure Anthropic Claude models
- Optional Add-ons: Enable Slack notifications and LLM Firewall as needed
Smart Service Management
- Automatically starts Docker services on launch
- Gracefully shuts down services on exit
- Profile-based optional services that can be toggled on/off
- Service restart prompts when configuration changes