The Agent Skills Directory

[SAFE]: The skill performs routine configuration tasks such as identifying available models and writing preference settings to ~/.agents/pstack-models.md.
[PERSISTENCE]: The skill manages a persistent configuration file to store model selection preferences across sessions, which is the intended functional behavior of the utility.
[INDIRECT_PROMPT_INJECTION]: The skill processes environment data and existing configuration files. Ingestion points include the orchestrator_capabilities output and the pstack-models.md file. While explicit boundary markers are not defined in the instructions, the skill capability inventory is limited to configuration writes and internal command invocation. Sanitization is performed by validating model candidates against a live catalog before configuration updates.

setup-pstack