The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a bash command block designed to be executed by the agent to check for the existence of an environment variable $POWERSUIT_OVERLAY and list contents of a directory using ls -1.
[DYNAMIC_EXECUTION]: The skill instructs the agent to dynamically load and follow additional instructions from computed paths based on the $POWERSUIT_OVERLAY environment variable, specifically directing the agent to prioritize these 'overlay' files over its own core instructions.
[EXTERNAL_DOWNLOADS]: The skill references external URLs on redd.it and machinebehavior.io to provide background documentation and historical context.
[INDIRECT_PROMPT_INJECTION]: The skill provides a framework for processing and scoring external claims, which creates a surface for indirect prompt injection where malicious data could influence the agent's logic.
Ingestion points: External 'substantive claims under discussion' processed in the gate logic (SKILL.md).
Boundary markers: None; the skill does not specify delimiters or warnings for the processed data.
Capability inventory: File system access (reading overlay README and directory files) and shell command execution (ls) as documented in SKILL.md.
Sanitization: No sanitization or escaping of external claims is performed before the scoring process.

label-gate