github

Warn

Audited by Gen Agent Trust Hub on Nov 4, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill is designed to execute a wide variety of shell commands using the gh (GitHub CLI) tool to interact with GitHub's infrastructure, including repository management, issue tracking, and workflow execution.
  • [DATA_EXFILTRATION]: The skill contains instructions for accessing sensitive local file paths. Specifically, it includes a command to add an SSH key to a GitHub account by reading from ~/.ssh/id_ed25519.pub and another to add GPG keys from local files.
  • [REMOTE_CODE_EXECUTION]: The skill documentation includes examples for installing external GitHub CLI extensions using gh extension install. This represents a mechanism for downloading and executing third-party code from remote repositories.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from external sources, creating a surface for indirect prompt injection attacks.
  • Ingestion points: The agent retrieves potentially attacker-controlled content through commands like gh issue view, gh pr view, gh gist view, and gh run view --log (which displays CI/CD logs).
  • Boundary markers: There are no instructions provided to the agent to use delimiters or ignore instructions embedded within the data retrieved from GitHub.
  • Capability inventory: The skill has broad capabilities, including writing to repositories, merging code, managing secrets, and installing extensions.
  • Sanitization: The instructions do not mention sanitizing, filtering, or validating the content retrieved from external sources before the agent processes it.
  • [DYNAMIC_EXECUTION]: By enabling the installation of extensions, the skill introduces the ability to execute dynamic code whose behavior is defined by the installed extension at runtime.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 4, 2026, 09:50 AM