NYC

pulumi-automation-api

Pass

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: SAFE
Full Analysis
  • [Prompt Injection] (SAFE): No patterns of prompt injection, such as instructions to ignore safety filters or override agent behavior, were detected.
  • [Data Exposure & Exfiltration] (SAFE): The skill does not access sensitive system files or exfiltrate data. Code examples use placeholders for configuration and demonstrate the use of the 'secret: true' flag to ensure sensitive data is encrypted in the Pulumi state.
  • [Indirect Prompt Injection] (SAFE): While the skill explains how to build self-service platforms that process external inputs (e.g., Slack bots), it serves as an educational resource for a legitimate SDK. Users building such platforms should implement their own input validation. Evidence Chain: 1. Ingestion points: Examples mention reading from 'deploy-config.json' and hint at external triggers like Slack. 2. Boundary markers: Not present in the simplified code snippets. 3. Capability inventory: The Pulumi Automation API provides high-privilege infrastructure management capabilities (creating/deleting cloud resources). 4. Sanitization: Not explicitly shown in the basic examples.
  • [Remote Code Execution] (SAFE): The skill describes the use of a programmatic infrastructure-as-code API. No malicious remote script downloads, piped command executions, or unverifiable dependencies were found.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 19, 2026, 03:29 PM