80-20-review

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and process untrusted data (source code and pull request content) provided by users or external systems.
  • Ingestion points: Processes code and PR descriptions when triggered by keywords like "review", "PR review", or "blast radius" (SKILL.md).
  • Boundary markers: None. There are no explicit instructions to the agent to ignore instructions embedded within the code or comments being reviewed.
  • Capability inventory: Uses several MCP tools for code analysis, including find_references, get_diagnostics, detect_antipatterns, and get_test_coverage_map (SKILL.md).
  • Sanitization: None. The skill relies on the output of static analysis tools and the LLM's own parsing logic.
  • [SAFE]: The skill's primary focus is defensive security. It explicitly instructs the agent to check for SQL injection, hardcoded secrets, authentication gaps, and improper error handling in .NET applications.
  • [SAFE]: The referenced MCP tools (detect_antipatterns, get_project_graph, etc.) are standard static analysis utilities for the .NET ecosystem and are used here for read-only diagnostics.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 06:49 PM
Security Audit — agent-trust-hub — 80-20-review