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

rpi-review

Pass

Audited by Gen Agent Trust Hub on Sep 21, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection Surface: The skill is designed to ingest and analyze external artifacts, including implementation evidence, plan critiques, and change records. This creates a potential surface where instructions embedded within the processed data could attempt to influence the agent's behavior.
  • Ingestion points: The skill resolves and reads task artifacts, changes records, and implementation evidence from the workspace tracking directory (Flow Steps 1 and 7).
  • Boundary markers: The skill utilizes structured templates (templates/review-log.md) and a defined protocol (references/review.md) to format findings, though it does not explicitly define sanitization steps for the input content.
  • Capability inventory: The skill possesses the ability to read project files, dispatch subagents/workers for specialized review tasks, and write review logs to the .copilot-tracking/ directory.
  • Sanitization: The skill implements a 'Parent-Worker' architecture where a primary agent (Parent) evaluates the findings of a subagent (Worker) and manages final decisions and user walkthroughs, providing a human-in-the-loop and multi-agent review layer.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 21, 2026, 08:44 PM
Security Audit — agent-trust-hub — rpi-review