kstack-interface-first-development
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes standard Git commands (
git rev-parse,git branch,git fetch,git merge,git pull) to verify the environment, manage implementation branches, and perform integrations. - [COMMAND_EXECUTION]: Employs the
wtCLI tool to create and manage Git worktrees, providing isolated environments for child sessions to work on specific tasks. - [INDIRECT_PROMPT_INJECTION]: The skill features an attack surface where the root session processes status updates, questions, and completion reports from child sessions via the
intercomtool. This risk is mitigated by explicit instructions for the root to manually verify integration evidence (commits/PRs) and seek user confirmation for all critical decisions and approvals. - [DYNAMIC_EXECUTION]: The root session dynamically assembles a 'handoff prompt' based on user-agreed designs. This prompt acts as the instruction set for child sessions. While this involves generating instructions at runtime, it is a core functional requirement for the multi-agent workflow and is guided by static templates.
- [SAFE]: Implements a robust 'Root/Child' architecture with clear separation of concerns. The root session acts as a security gatekeeper, preventing child sessions from directly communicating with the user for reviews or performing unverified repository integrations.
Audit Metadata