The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is explicitly designed to run shell commands to verify game builds. It instructs the agent to execute an "authority verify" command and a "signature_command" found within the project context. This creates a surface for arbitrary command execution on the host environment.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from multiple external sources including PRODUCT_BRIEF.md, BUILD_BRIEF, and GAME_DESIGN. These files are used to determine which commands the agent should execute and what assertions to make.
Ingestion points: PRODUCT_BRIEF.md, BUILD_BRIEF, and GAME_DESIGN files (specified in SKILL.md).
Boundary markers: Absent; the instructions do not specify any delimiters or warnings to ignore instructions embedded within these project files.
Capability inventory: Shell command execution (via the verify command and signature_command mentioned in SKILL.md step 4).
Sanitization: Absent; the skill does not require validation or escaping of the commands retrieved from the project metadata before execution.
[DYNAMIC_EXECUTION]: The skill utilizes dynamic command assembly and execution. It relies on finding a signature_command or a specific verification command within the project's metadata at runtime and executing it to gather evidence for the QA report.

game-qa