The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing external data from the local project repository to influence its code generation logic.
Ingestion points: The skill reads content from existing test files (tests/**/*_test.*), design documents (design/gdd/*.md), and architecture registries (docs/architecture/tr-registry.yaml) to extract patterns.
Boundary markers: No explicit delimiters or instructions are provided to the agent to differentiate between data and potentially malicious instructions embedded within the project files.
Capability inventory: The skill uses Read, Glob, Grep, and Write tools to inspect the environment and create new helper scripts in the tests/helpers/ directory.
Sanitization: The instructions do not define any sanitization, escaping, or validation steps for the content extracted from project files before it is interpolated into the generated code templates.
[REMOTE_CODE_EXECUTION]: The skill performs dynamic script generation based on project context and predefined templates.
Evidence: The skill generates executable source code for Godot (GDScript), Unity (C#), and Unreal Engine (C++). While the agent does not execute the generated code itself, it writes these files to the local file system using the Write tool.

test-helpers