running-github-actions-efficiently

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions to download binary releases for SwiftLint and SwiftFormat from external GitHub repositories (realm/SwiftLint and nicklockwood/SwiftFormat).
  • [PRIVILEGE_ESCALATION]: The provided shell scripts use sudo install to move downloaded binaries into /usr/local/bin/, which requires root privileges on the runner environment.
  • [COMMAND_EXECUTION]: The instructions include multiple shell command blocks for environment setup, tool installation, and running linters (curl, unzip, install, swiftlint, swiftformat).
  • [INDIRECT_PROMPT_INJECTION]: The skill includes self-referential commentary targeting the repository it is placed in, claiming to have identified missing configurations in .github/workflows/ci.yml.
  • Ingestion points: SKILL.md (specifically the 'Note for this repository' section).
  • Boundary markers: Absent; the text is presented as authoritative guidance.
  • Capability inventory: The skill provides instructions for modifying CI workflows and executing shell commands.
  • Sanitization: None; the text is processed as standard documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 11:02 AM
Security Audit — agent-trust-hub — running-github-actions-efficiently