setup-pstack
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill attempts to identify available AI models by querying a system CLI or a models API at runtime.
- [DATA_EXPOSURE]: The skill reads from and writes to the local file system, specifically targeting the configuration path
~/.cursor/rules/pstack-models.mdcto manage persistent settings. - [INDIRECT_PROMPT_INJECTION]: The skill produces an 'always-applied' instruction file (
.mdc). This file acts as a persistent prompt that governs the agent's model selection logic in future sessions. The ingestion point for this instruction set is the confirmed model slugs detected from the environment, which are then written into a structured markdown configuration. - [DYNAMIC_EXECUTION]: The skill generates a configuration file that contains logic-influencing instructions for the AI agent. This file is written at runtime and interpreted by the agent system in subsequent sessions to determine subagent behavior.
Audit Metadata