The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's primary function is to 'Run and capture' technical experiments, which explicitly requires the agent to execute shell commands and scripts to observe system behavior, performance, or tool integration results.
[DYNAMIC_EXECUTION]: The core workflow involves creating the 'smallest runnable experiment' and building 'artifacts' that are executed at runtime to validate or invalidate technical hypotheses.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process technical questions, environmental conditions, and hypotheses provided by the user or external sources to generate experimental code. This creates a surface area for indirect injection where the input could influence the generated commands.
Ingestion points: Technical questions and environment descriptions provided during the 'Frame the uncertainty' step in SKILL.md.
Boundary markers: The instructions emphasize framing a specific observable question, which serves as a logical boundary, though no explicit string delimiters are mandated.
Capability inventory: The skill facilitates file creation, command execution, and measurement capturing across the host environment as defined in SKILL.md.
Sanitization: The instructions do not explicitly detail sanitization steps, relying on the agent's internal logic to build the probe safely.

spike