The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local script at ../rex-subagents/scripts/subagent to probe for available AI models and their reasoning capabilities. This is a core part of its model detection functionality.
[DYNAMIC_EXECUTION]: The skill can conditionally load and follow instructions from other local instruction files (../create-verification-skill/SKILL.md and ../create-formal-verification-skill/SKILL.md) to help the user generate project-specific verification tools.
[INDIRECT_PROMPT_INJECTION]: The skill reads and processes an existing configuration file at ~/.config/pstack/pstack-models.md. It mitigates potential injection risks by validating all choices against the set of models detected in the environment.

setup-pstack