analyze-ci-failures

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • Command Execution: The skill uses PowerShell cmdlets like Invoke-RestMethod and Invoke-WebRequest to automate the retrieval of build data. This is a standard automation practice for CI/CD diagnostics within the vcpkg project ecosystem.
  • External Downloads: The workflow involves downloading build artifacts and logs from dev.azure.com and visualstudio.com. These are official, well-known services managed by the project vendor for hosting CI results.
  • Indirect Data Processing: The skill is designed to ingest and parse build logs to identify failure patterns. As these logs are sourced externally, this represents a data processing surface common to analysis tools, which the skill manages using structured templates.
  • Instructional Guardrails: The skill includes explicit usage constraints and critical rules in SKILL.md to ensure the agent remains focused on diagnostic tasks and avoids modifying sensitive project files.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 05:23 AM
Security Audit — agent-trust-hub — analyze-ci-failures