infra-build

Pass

Audited by Gen Agent Trust Hub on Aug 5, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to use standard CLI tools including gh, terraform, fly, and make to manage infrastructure resources. These actions are performed within the scope of user-approved design policies.
  • [DATA_EXPOSURE]: The agent reads local project context and configuration files. It is explicitly warned not to write sensitive credentials or secrets into the chat or files, instead providing only the necessary commands for the user to register them securely.
  • [PROMPT_INJECTION]: The skill ingests data from GitHub issues, which serves as a potential surface for indirect prompt injection. This risk is addressed through specific design choices: 1. Ingestion points: External content is read using gh issue view. 2. Boundary markers: A mandatory confirmation step (Step 4) requires the agent to present a summary for human approval before execution. 3. Capability inventory: The agent can write infrastructure code and run deployment commands. 4. Sanitization: Instructions mandate that secrets be handled via registration procedures rather than direct output.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 5, 2026, 07:21 AM
Security Audit — agent-trust-hub — infra-build