skills/microsoft/rego-cpp/finalize-pr/Gen Agent Trust Hub

finalize-pr

Pass

Audited by Gen Agent Trust Hub on Aug 19, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • Command Execution: The skill utilizes several standard development tools to prepare a branch for release. This includes repository management with git, build and test automation with ninja and ctest, and searching for version strings with grep. These are expected actions for a PR finalization workflow.
  • External Downloads: The skill performs network operations to synchronize with remote repositories and update dependencies.
  • It uses git fetch to retrieve updates from the primary repository.
  • It uses cargo update to refresh Rust dependencies from the official package registry.
  • These operations target well-known, established services as part of standard software maintenance.
  • Indirect Prompt Injection Surface: The skill processes external, potentially untrusted data which could theoretically be used to influence the AI's behavior.
  • Ingestion points: The skill reads branch diffs, commit logs, and code comments (SKILL.md).
  • Boundary markers: No explicit delimiters or instruction-ignore warnings are implemented for the ingested data.
  • Capability inventory: The skill can execute local commands (git, ninja, ctest, cargo, grep) and write to files within the repository (VERSION, CHANGELOG, etc.).
  • Sanitization: The skill does not explicitly sanitize the ingested text before using it to generate summaries or documentation. This is a common design pattern for developer-oriented tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 19, 2026, 11:42 PM
Security Audit — agent-trust-hub — finalize-pr