The Agent Skills Directory

[COMMAND_EXECUTION]: Executes shell commands for infrastructure lifecycle management using terraform and version control operations via git.
[COMMAND_EXECUTION]: Invokes third-party CLI tools codex and gemini to perform automated analysis on local project files.
[DATA_EXFILTRATION]: Transmits local repository data and version control diffs to well-known external AI service providers for review purposes.
[PROMPT_INJECTION]: The skill processes untrusted input from local code files and git history, which is subsequently provided as context to other AI models, presenting an indirect prompt injection risk.
Ingestion points: git diff output and local Terraform configuration files (.tf) located in vendor-specific paths.
Boundary markers: Absent; code content is interpolated directly into command strings for AI tools without delimiters or instructions to ignore embedded commands.
Capability inventory: Full shell access to execute infrastructure-altering terraform commands and interaction with remote AI APIs.
Sanitization: No evidence of sanitization or filtering of the source code content before it is processed by the AI models.

test-changes