The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests and processes potentially untrusted external data, including Terraform HCL files, modules, variable files, and plan outputs. An attacker could embed malicious instructions within code comments or metadata in these files to influence the agent's behavior. The skill possesses capabilities to write to the filesystem and execute shell commands, creating a pathway for these instructions to be acted upon.\n
Ingestion points: The skill reads .tf files, modules, .tfvars, and Terraform plan outputs during the ACQUIRE scene and Analyze step (SKILL.md, resources/execution-protocol.md).\n
Boundary markers: No explicit delimiters or instructions to ignore embedded commands were identified within the prompt templates.\n
Capability inventory: The skill can write Terraform files, modules, and documentation, and execute terraform CLI commands (fmt, validate, plan) and security scanners (checkov, tfsec) (SKILL.md, resources/execution-protocol.md).\n
Sanitization: No specific sanitization or filtering logic is implemented for the content of the ingested infrastructure files.\n- [COMMAND_EXECUTION]: The skill is designed to execute local system commands via the Terraform CLI and various security scanning tools like checkov and tfsec to validate and plan infrastructure changes. While these are legitimate operational tools for the skill's purpose, they represent a capability that requires user oversight during execution (SKILL.md, resources/execution-protocol.md).

oma-tf-infra