terraform
Warn
Audited by Gen Agent Trust Hub on Mar 23, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill is instructed to access highly sensitive local configuration files, including AWS credentials, gcloud configurations, and Azure CLI settings. While intended for cloud provider detection, these files contain long-term secrets and access tokens that could be exposed in the agent's context or output.
- [PROMPT_INJECTION]: The instructions utilize a strong behavioral override by mandating an "AUTONOMOUS MODE" and explicitly ordering the agent to "Do NOT ask questions" and "Do NOT pause for confirmation." This reduces the user's opportunity to review sensitive operations such as credential harvesting or file system modifications.
- [PROMPT_INJECTION]: The skill has a significant indirect prompt injection surface as it ingests untrusted data from various project files to drive its generation logic.
- Ingestion points: Reads content from
Dockerfile,prisma/schema.prisma,serverless.yml, and existing.tffiles. - Boundary markers: Absent. The instructions do not define delimiters or provide warnings to the agent to disregard embedded instructions within these external files.
- Capability inventory: The skill possesses full read/write access to the workspace to scaffold the Terraform project structure and telemetry logging.
- Sanitization: Absent. There is no evidence of sanitization, escaping, or validation of the data read from project files before it is interpolated into the generation templates.
Audit Metadata