architect

Pass

Audited by Gen Agent Trust Hub on Oct 12, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill implements a comprehensive design workflow using a multi-phase approach (Ground, Sketch, Agree, Implement, Scrap). It accesses application-specific configuration files (e.g., ~/.config/pstack/pstack-models.md) and state files (e.g., ~/.local/state/pstack/checklists/) to manage the development process.
  • [INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface as it processes external repository code and user-defined tasks to generate design artifacts. Ingestion points include repository subsystems and checklist files; boundary markers are provided via structured templates and discrete phases; capabilities include executing sub-skills (how, arena, why) and writing design rationales; no explicit sanitization logic is described. Given its primary purpose as a development aid, this surface is a standard operational requirement and does not present a high risk.
  • [SAFE]: No evidence of credential harvesting, unauthorized network access, or malicious code execution was found. The use of parallel runners for design exploration includes mentions of process isolation via git worktrees or subdirectories.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 12, 2026, 04:02 AM