run-self-service

Run Self-Service Action Skill

Execute Port self-service actions through natural language interface.

When to Use

• User wants to create cloud infrastructure
• Provisioning a new service or environment
• Triggering any Port self-service action

Execution Process

Step 1: Discover Available Actions

list_actions()

Find actions matching user intent.

Step 2: Get Action Schema

get_action(identifier="action-name")

Retrieve:

• Required inputs
• Input types and validation
• Approval requirements

Step 3: Collect User Inputs

For each required input, prompt the user:

• Use the input's title and description
• Show enum options if available
• Validate format (email, URL, etc.)

Present like a form:

📝 Create New Service

Service Name: [required]
> 

Team: [required, select one]
  • platform-team
  • payments-team  
  • growth-team

Environment: [required, select one]
  • development
  • staging
  • production

Description: [optional]
> 

Step 4: Validate Inputs

Before submission:

• Check required fields
• Validate formats
• Confirm with user

Step 5: Execute Action

run_action(identifier="action-name", properties={
  "service_name": "user-input",
  "team": "selected-team",
  "environment": "selected-env"
})

Step 6: Monitor Execution

If action is async:

get_action_run(run_id="run-123")

Report status: pending → running → success/failure

Example Interactions

User: "I need a new S3 bucket"

Agent:

1. Finds provision_s3_bucket action
2. Gets schema: bucket_name, region, access_level
3. Prompts:

   📝 Provision S3 Bucket
   
   Bucket Name: [required]
   > 
   
   Region: [required, select one]
     • us-east-1
     • eu-west-1
   
   Access Level: [required, select one]
     • private
     • public-read
   

4. Executes action
5. Reports: "✅ Bucket created: my-bucket.s3.amazonaws.com"

User: "Create a new microservice"

Agent:

1. Finds scaffold_service action
2. Collects: name, language, team, repo visibility
3. Executes
4. Reports: "✅ Service created with repo, CI pipeline, and Port entity"

Form-Like Experience

When collecting inputs:

• Show one input at a time for clarity
• Provide defaults where available
• Show validation errors inline
• Confirm before execution

Quality Checklist

• Found correct action for user intent
• Retrieved complete input schema
• Collected all required inputs
• Validated input formats
• Confirmed before execution
• Reported execution status