Seeds
AI personas for different cloud tasks
Seeds
Seeds are AI personas — specialized system prompts that give NubiferAI deep expertise in specific cloud domains. Each seed tells the AI how to think, what tools to use, and what format to produce output in.
How Seeds Work
When you submit a prompt, the active seed provides:
- System prompt — Instructions that shape the AI's behavior and expertise
- Tool definitions — What actions the AI can take (create resources, run commands, etc.)
- Output formats — What export formats are supported (Terraform, shell, CloudFormation, etc.)
- Context — Documentation references, software dependencies, and domain knowledge
Browsing Seeds
GUI
Click Seeds in the header bar or press Ctrl+B to open the seed browser:
- Search by name or description
- Filter by tags (aws, terraform, kubernetes, etc.)
- Click a card to see full details
- Click Use This Seed to pre-select it for your next prompt
CLI
# List all seeds
nubiferai seed list
# Filter by tag
nubiferai seed list --tag aws
nubiferai seed list --tag finops
# Show full details
nubiferai seed show aws-deployer
Available Seeds
AWS (12 seeds)
| Seed | Description | Key Tags |
|---|---|---|
aws-deployer | General AWS deployment with Terraform | aws, terraform, iac |
aws-auditor | Security auditing and compliance checks | aws, security |
terraform-expert | Deep Terraform expertise | aws, terraform |
kubernetes-expert | Kubernetes and Helm deployments | aws, kubernetes |
aws-solutions-architect | Architecture design and review | aws, architecture |
aws-developer | Serverless, containers, and CI/CD | aws, serverless |
aws-security | IAM, encryption, and security review | aws, security, iam |
aws-lambda | Lambda function deployment | aws, serverless |
aws-ecs-eks | Container orchestration (ECS/EKS) | aws, containers |
aws-rds | Database setup and management | aws, database |
aws-networking | VPC, subnets, and network design | aws, networking |
aws-iam | IAM policies, roles, and access design | aws, iam |
aws-s3 | S3 bucket configuration | aws, storage |
aws-cloudfront | CDN and DNS setup | aws, cdn |
aws-dynamodb | DynamoDB table design | aws, database |
Google Cloud (5 seeds)
| Seed | Description | Key Tags |
|---|---|---|
gcp-gke | Google Kubernetes Engine deployment | gcp, kubernetes |
gcp-cloud-run | Cloud Run serverless containers | gcp, serverless |
gcp-data | BigQuery and data infrastructure | gcp, database |
gcp-networking | VPC and network design | gcp, networking |
gcp-iam | IAM and security configuration | gcp, iam |
Azure (5 seeds)
| Seed | Description | Key Tags |
|---|---|---|
azure-aks | Azure Kubernetes Service | azure, kubernetes |
azure-app-service | App Service deployment | azure, serverless |
azure-database | SQL and Cosmos DB setup | azure, database |
azure-networking | VNet and network design | azure, networking |
azure-iam | Azure AD and role assignments | azure, iam |
FinOps (4 seeds)
| Seed | Description | Key Tags |
|---|---|---|
cost-optimizer | Cost reduction recommendations | aws, finops |
cost-analyzer | Cost Explorer and billing analysis | aws, gcp, azure, finops |
cost-estimator | Pre-deployment cost estimation | aws, gcp, azure, finops |
budget-manager | Budget alerts and governance | aws, gcp, azure, finops |
DevOps (4 seeds)
| Seed | Description | Key Tags |
|---|---|---|
cicd-expert | CI/CD pipeline design | devops, cicd |
monitoring-expert | Observability and alerting | devops, monitoring |
docker-expert | Dockerfile and container builds | devops, containers |
ansible-expert | Ansible playbook generation | devops, iac |
Security & Compliance (1 seed)
| Seed | Description | Key Tags |
|---|---|---|
compliance-expert | Compliance frameworks and governance | security, compliance |
General (1 seed)
| Seed | Description | Key Tags |
|---|---|---|
general | General cloud questions and advice | aws, general |
Using Seeds
Pre-selecting a Seed
GUI: Click a Quick Action button, or browse Seeds and click Use This Seed.
CLI:
nubiferai nucleate --seed terraform-expert "Refactor my VPC module"
Default Seed
If no seed is explicitly selected, NubiferAI uses your configured default (typically aws-deployer).
Change the default:
GUI: Settings > General > Default Seed
CLI:
# In ~/.config/nubiferai/config.toml
default_seed = "terraform-expert"
Seed Anatomy
Each seed is a TOML file with these sections:
id = "aws-deployer"
name = "AWS Deployer"
icon = "🏗"
description = "Deploy infrastructure to AWS using Terraform"
version = "1.0"
author = "NubiferAI"
system_prompt = "You are an expert AWS infrastructure engineer..."
tags = ["aws", "terraform", "iac", "devops"]
[outputs]
formats = ["terraform", "shell", "json", "cfn"]
executors = ["terraform"]
[ui]
quick_action_label = "Deploy to AWS"
quick_action_position = 1
[[docs]]
name = "Terraform AWS Provider"
url = "https://registry.terraform.io/providers/hashicorp/aws/latest/docs"
[[software]]
name = "terraform"
version = "1.5+"
required = true
Creating Custom Seeds
# Scaffold a new seed
nubiferai seed new my-custom-seed
# Edit the generated TOML file
# Place in ~/.config/nubiferai/seeds/ for personal use
# Or in your workspace's .nubiferai/seeds/ for team sharing
Custom seeds are loaded alongside built-in seeds and appear in the GUI and CLI.
Seed Search Order
- Workspace seeds (
.nubiferai/seeds/) - User seeds (
~/.config/nubiferai/seeds/) - Built-in seeds (bundled with NubiferAI)
User/workspace seeds with the same ID override built-in seeds.