gh-cli-flow

Pass

Audited by Gen Agent Trust Hub on Feb 23, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill facilitates the ingestion and processing of untrusted data from GitHub pull requests, issues, and review comments, creating a surface for indirect prompt injection.
  • Ingestion points: Operations in SKILL.md (e.g., gh pr view, gh issue view), references/api-readonly.md (REST API calls for comments), and references/pr-comment-workflow.md (GraphQL queries for review threads) retrieve external, attacker-controllable text.
  • Boundary markers: The instructions do not specify the use of delimiters or boundary markers to isolate retrieved data from the agent's internal instruction set.
  • Capability inventory: The skill possesses significant write capabilities, including merging PRs (gh pr merge), approving reviews (gh pr review --approve), and creating issues/PRs. Additionally, references/line-comments.md provides patterns for modifying or deleting review comments via gh api.
  • Sanitization: No evidence of sanitization, filtering, or validation of the retrieved content is present in the skill's logic.
  • [COMMAND_EXECUTION]: The skill relies extensively on executing shell commands via the gh (GitHub CLI) tool.
  • Evidence: SKILL.md and the reference files provide numerous templates for command execution to manage repositories, CI/CD runs, and workflows. While this is the primary purpose of the skill, it involves high-volume interaction with a local binary tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 23, 2026, 10:10 PM