The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes local CLI tools terraform (init, fmt, validate) and coder (templates push, versions list) to manage workspace configurations.
Evidence: The workflow utilizes terraform init, terraform fmt, and terraform validate to ensure the integrity of the modified main.tf file.
Evidence: The coder templates push command is used to deploy the updated templates to the Coder platform.
[EXTERNAL_DOWNLOADS]: Infrastructure snippets and documentation are retrieved from the official Coder registry and documentation site.
Evidence: Fetches Terraform module definitions and README content from https://registry.coder.com and https://coder.com/docs.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from the module registry which is then used to modify project files, creating a potential surface for indirect instruction injection if upstream sources were compromised.
Ingestion points: Remote Terraform snippets and README files retrieved from registry.coder.com.
Boundary markers: The skill requires an explicit user 'yes/no' confirmation after showing the proposed code snippet before any file modification occurs.
Capability inventory: Possesses capabilities to write to main.tf, execute shell commands for validation, and push code to the Coder deployment.
Sanitization: No specific sanitization or filtering is applied to the Terraform snippets beyond standard Terraform validation commands.

modules