{
  "context": "Tests whether the agent uses the gh CLI correctly for all GitHub operations, uses --body-file for PR body content, sets explicit --base and --head flags when opening a PR, waits for CI after opening the PR, and never includes AI co-authorship attributions. The scenario evaluates the octocat skill's specific workflow preferences.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "gh CLI only",
      "description": "All GitHub operations are performed using the `gh` CLI. No instructions suggest using the GitHub web interface, REST API via curl, or other GitHub clients.",
      "max_score": 10
    },
    {
      "name": "--body-file flag",
      "description": "The `gh pr create` command uses `--body-file <path>` instead of `--body \"...\"` for the PR description. The PR body content is written to a temporary file first.",
      "max_score": 12
    },
    {
      "name": "Explicit --base and --head",
      "description": "The `gh pr create` command includes explicit `--base <branch>` and `--head <branch>` flags, not relying on defaults.",
      "max_score": 10
    },
    {
      "name": "CI check wait command",
      "description": "After opening the PR, the script or instructions include `gh pr checks <num> --watch` (or equivalent with `2>&1`) to wait for CI to complete.",
      "max_score": 12
    },
    {
      "name": "No AI co-authorship",
      "description": "No commit message, PR body, or script output contains 'Co-Authored-By: Claude', 'Co-Authored-By: Codex', 'Generated by ChatGPT', or any similar AI attribution string.",
      "max_score": 14
    },
    {
      "name": "No signing config changes",
      "description": "The script does NOT modify git signing configuration (`user.signingkey`, `gpg.program`, `commit.gpgsign`) in any git config command.",
      "max_score": 10
    },
    {
      "name": "Structured PR body content",
      "description": "The PR body written to the temporary file is structured markdown with at least a summary/description section (not a single line of plain text).",
      "max_score": 8
    },
    {
      "name": "Shell script artifact",
      "description": "The output includes a shell script (`.sh`) that automates the described PR workflow, so the process is reproducible and inspectable.",
      "max_score": 8
    },
    {
      "name": "Rebase validation step",
      "description": "If interactive rebase is performed, the script or instructions include `git log --oneline -n 10` (or similar) to verify history after the rebase.",
      "max_score": 8
    },
    {
      "name": "Temp file cleanup",
      "description": "Any temporary files created for the PR body (e.g., `/tmp/pr-body.md`) are removed after use, or the script is structured so they don't persist.",
      "max_score": 8
    }
  ]
}