Enterprise-grade AIOps & LLMOps automation agent built for DevOps teams.
PicoClaw FinOps Copilot is a Slack-native, AI-assisted cloud cost intelligence platform that integrates AWS Cost Explorer with deterministic cost analysis and LLM-based executive reporting.
It bridges the gap between DevOps, FinOps, LLMOps, and AIOps by helping engineering teams:
- Compare monthly cloud spend
- Detect cost anomalies and idle resources
- Automate scheduled reports
- Explain cost changes in plain language
- Operate securely using IAM roles (no static credentials)
- π Numbers are deterministic (computed via AWS SDK)
- π§ AI is used only for explanation and summarization
This prevents hallucination and ensures financial accuracy.
This repository comes with a built-in AWS FinOps skill for PicoClaw, allowing it to act as your autonomous cost analyst.
Capabilities:
- Cost Summary / Service Breakdown: Get top 5 service costs for any time range (e.g., "last 7 days").
- Monthly Cost Comparison: Compare current month spend vs previous month spend with percentage change highlighting.
- Idle Resource Detection: Finds unused EBS volumes (
availablestate), calculates monthly waste, and offers cleanup assistance safely.
- Daily cost summary
- Service-level cost spike detection
- Weekly executive report
- Plain-language explanations of cost increases
- Slack-friendly formatting (no massive raw JSON dumps)
- Safe read-only executions (destructions require explicit user confirmation)
- IAM Role-based authentication
- No AWS access keys stored
- Read-only Cost Explorer access
- No destructive operations without approval
- Slack token-based access control
- Ubuntu EC2 instance (or compatible Linux server)
- IAM Role with Cost Explorer permissions
- AWS CLI v2 installed
- Slack App (Socket Mode enabled)
- NVIDIA-hosted LLM (or compatible OpenAI-style endpoint)
- PicoClaw installed
Minimum required permissions:
ce:GetCostAndUsageec2:DescribeInstancesec2:DescribeVolumessts:GetCallerIdentity
Verify:
aws sts get-caller-identityUpdate ~/.picoclaw/config.json:
{
"agents": {
"defaults": {
"provider": "moonshot",
"model": "moonshotai/kimi-k2.5"
}
}
}Enable Slack channel:
"slack": {
"enabled": true,
"bot_token": "xoxb-...",
"app_token": "xapp-...",
"allow_from": []
}Enable Telegram channel (Alternative):
"telegram": {
"enabled": true,
"bot_token": "123456:ABC-DEF1234ghIkl-zyx57W2v1u123ew11",
"allow_from": ["123456789"]
}picoclaw gatewayHealth endpoints available:
http://127.0.0.1:18790/health
http://127.0.0.1:18790/ready
Here are examples of what the FinOps Copilot outputs when you ask it for cost data in Slack or Telegram:
| Automation | Description |
|---|---|
| Daily Cost Summary | 9 AM automated report |
| Budget Alert | Notify when spend exceeds threshold |
| Weekly Executive Report | Slack-ready cost summary |
| Anomaly Detection | Detect sudden cost spikes |
| Waste Detection | Identify idle resources |
| Phase | Capability | Status |
|---|---|---|
| Phase 1 | Slack FinOps Integration | β |
| Phase 2 | Deterministic Cost Engine | π§ |
| Phase 3 | Automated Alerts | π§ |
| Phase 4 | Anomaly Detection | β³ |
| Phase 5 | Autonomous Optimization | β³ |
Full roadmap available in ROADMAP.md.
- IAM Role only (no stored AWS keys)
- Read-only billing access
- Slack event filtering
- No destructive changes without confirmation
- Workspace execution isolation
- Audit logging enabled
Planned enhancements:
- AWS SDK integration (remove CLI dependency)
- Multi-account AWS Organizations support
- Slack approval workflow
- Terraform auto-deployment
- CI pipeline with lint + security scan
- Structured JSON logging
- Event deduplication guard
Most DevOps teams lack:
- Real-time cost visibility
- Slack-integrated FinOps tooling
- Automated anomaly detection
- AI-assisted cost explanations
This project bridges:
- DevOps
- FinOps
- LLMOps
- Cloud Automation
It demonstrates real-world distributed systems engineering combined with AI orchestration.
Build a fully autonomous cloud FinOps engineer that:
- Continuously monitors AWS spend
- Detects waste automatically
- Explains cost shifts clearly
- Suggests optimizations
- Operates safely under IAM constraints
- Integrates natively with team workflows
MIT License



