terraform-patterns
Fail
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: HIGHCREDENTIALS_UNSAFECOMMAND_EXECUTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The file
examples/main.tfcontains several instances of hardcoded sensitive information. - Hardcoded AWS
access_key(AKIAIOSFODNN7EXAMPLE) andsecret_key(wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY) are present in the provider configuration block. - A plain-text database password (
admin123) is hardcoded in theaws_db_instanceresource. - Inline secrets are present in the
user_datascript for an EC2 instance, including aDB_PASSWORDand anAPI_KEY(sk-live-abc123def456). - [COMMAND_EXECUTION]: The
tf_module_analyzer.pyandtf_security_scanner.pyscripts execute as standalone Python processes and are intended to be run by the agent to perform local file analysis. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted user-provided Terraform code. While the scripts use regular expressions for parsing rather than LLM processing, the results are returned to the agent context.
- Ingestion points: Local
.tffiles specified via the--pathargument. - Capability inventory: Python scripts perform file system reads (
Path.read_text()) and directory walking (os.walk()). - Sanitization: The scripts use static regex patterns to identify misconfigurations; however, the agent's interpretation of the output could be influenced by malicious comments or resource names in the scanned code.
- Boundary markers: Not explicitly defined in the provided script outputs, though findings are structured as JSON or plain text.
Recommendations
- AI detected serious security threats
Audit Metadata