research-plan-implement
Pass
Audited by Gen Agent Trust Hub on Apr 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Orchestrates the workflow by spawning subagents via the Agent tool and calling localized skills such as 'rpikit:researching-codebase' and 'rpikit:writing-plans'.- [PROMPT_INJECTION]: Involves a potential surface for indirect prompt injection as research data from external sources is synthesized into implementation plans; however, the skill mitigates this through mandatory human-in-the-loop approval gates before code is modified.- [SAFE]: Implementation subagents utilize 'isolation: worktree', a security feature that contains changes within a temporary git environment to prevent unintended side effects on the primary branch.- [SAFE]: Adheres to the principle of least privilege for the orchestrator, keeping the main agent context 'thin' and delegating high-privilege tasks to specialized subagents with focused scopes.
Audit Metadata