skills/microsoft/hve-core/rpi-plan/Gen Agent Trust Hub

rpi-plan

Pass

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection Surface: The skill transforms research and requirements into planning artifacts that may contain shell commands for validation, creating a potential path for untrusted data to influence generated commands.
  • Ingestion points: Research artifacts located in .copilot-tracking/research/ and user-provided requirements enter the agent context in the first phase.
  • Boundary markers: The skill utilizes structured markdown templates for data separation; however, it does not currently employ explicit 'ignore embedded instructions' delimiters for the interpolated research content.
  • Capability inventory: The skill involves file writing to the project's tracking directory and generating command-line instructions for build and test phases. It also has the ability to dispatch tasks to external subagents for research and validation.
  • Sanitization: The process includes a mandatory 'Plan Validator' step in Phase 3. This check is designed to identify and resolve findings in the generated plans and details before any implementation handoff is permitted.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 30, 2026, 09:46 AM
Security Audit — agent-trust-hub — rpi-plan