The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill executes a shell script (yaml-helper.sh) at load time using the ! syntax to resolve configuration. The script is located in a directory relative to the skill (../../hooks/).- [PROMPT_INJECTION]: The instructions explicitly direct the agent to bypass standard safety protocols, specifically the requirement to ask for user confirmation before writing or editing files. It labels this a 'deliberate, bounded exception' and instructs subagents to write without asking.- [COMMAND_EXECUTION]: The skill automates file-writing operations by subagents without per-file user oversight, relying on phase-level gates instead of individual action confirmations.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input and reads external files, creating a surface where malicious instructions in that data could influence the agent's actions.
Ingestion points: User-provided season/event arguments and external files like design/live-ops/economy-rules.md and design/live-ops/ethics-policy.md.
Boundary markers: No explicit delimiters or 'ignore embedded instructions' warnings are provided when passing this data to subagents.
Capability inventory: Subagents have access to Write and Edit tools, and the orchestrator can execute restricted Bash commands.
Sanitization: No validation or sanitization of user-provided strings or file content is performed before interpolation into agent prompts.

team-live-ops