Aigent Squad

Multi-agent platform for AWS and Kubernetes operations.

One supervisor, five specialist agents, zero code to add a new one. Config-driven, read-only by default, fully observable.

---

What it does

Aigent Squad receives operational questions in natural language and routes them to the right specialist agent. Each agent collects read-only context from its data sources and synthesizes a structured answer using Claude on AWS Bedrock.

Your question
     │
     ▼
Supervisor (classifier + orchestrator)
     │
     ├──▶ AWS Agent       → EC2, RDS, S3, IAM, Cost Explorer
     ├──▶ Kubernetes Agent → Pods, deployments, events, metrics
     ├──▶ FinOps Agent    → Cost analysis, savings plans
     ├──▶ DevOps Agent    → CI/CD, GitLab, pipelines
     └──▶ Observability   → Metrics, logs, traces, alerts

---

Key properties

Property	Detail
Read-only	Agents never execute mutations — analysis and recommendations only
Config-driven	Add an agent with 2 files (agent.yaml + prompt.md), no code
Single image	Supervisor runs all agents in-process; one container to deploy
Cost attributed	Per-agent token counting via Bedrock inference profiles
Observable	OTel traces, RED metrics, and structured logs on every component
Conversation history	DynamoDB with 24h TTL per user/session

---

Quick install

helm repo add staffops https://staffops.github.io/helm-charts/
helm repo update

helm install aigent-squad staffops/aigent-squad \
  --namespace aigent-squad --create-namespace \
  --set redis.host=<your-elasticache-endpoint>

Getting Started → Architecture →

---

Source

github.com/StaffOps/staffops-aigent-squad — Apache 2.0