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

rpi-review

Pass

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • Local Command Execution: The skill is designed to discover and execute validation commands found within the project environment, such as those in a Makefile, package.json, or pyproject.toml. This allows the agent to verify the implementation using the project's own testing and build infrastructure as intended for the review process.
  • Processing Local Project Artifacts: The skill ingests data from local files, including implementation plans and changes logs located in the .copilot-tracking directory. This information is used to populate review templates and assess task completion.
  • Subagent Integration: The skill utilizes subagents for specialized validation tasks. This modular approach facilitates analysis of implementation quality while maintaining a structured review protocol.
  • Indirect Prompt Injection Surface: The skill processes various project files to synthesize findings, which is a common pattern for automated review tools.
  • Ingestion points: Project configuration files and subagent output files located in the .copilot-tracking/ directory.
  • Boundary markers: The skill employs structured markdown templates to separate different types of metadata and findings.
  • Capability inventory: The agent can execute shell commands found in the project configuration and invoke subagents for validation.
  • Sanitization: The instructions focus on summarizing findings and recording them into a predefined log format, minimizing the direct reflection of raw input into commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 30, 2026, 09:46 AM
Security Audit — agent-trust-hub — rpi-review