Update to Greptile

{
  "name": "review",
  "description": "Review the current working branch with the bundled Greptile CLI, optionally against a specified base branch or with review instructions.",
  "included_files": [
    {
      "relative_path": "scripts/greptile.mjs",
      "size_in_bytes": 10697706
    },
    {
      "relative_path": "scripts/greptile.version",
      "size_in_bytes": 6
    }
  ],
  "skill_md_contents": "---\nname: review\ndescription: Review the current working branch with the bundled Greptile CLI, optionally against a specified base branch or with review instructions.\n---\n\nRun a Greptile review from the user's repository. No open pull request is needed.\n\nResolve `<skill-root>` as the directory containing this `SKILL.md`.\nSubstitute that absolute path in the command; use the bundled script rather\nthan a separately installed `greptile` executable.\nKeep the working directory in the repository being reviewed.\n\n```sh\nGREPTILE_NO_AUTO_INSTALL=1 GREPTILE_NO_UPDATE_CHECK=1 node \"<skill-root>/scripts/greptile.mjs\" review --agent\n```\n\nIf the user specifies a base branch, append `--branch` with that branch. Pass\nany focus instructions verbatim with `--instructions`. Shell-quote both\nvalues as literal arguments so dollar signs, backticks, quotes, and command\nsubstitutions in user text cannot execute or change the text.\n\nAlways pass `--agent`. The environment variables suppress the Mermaid renderer\ndownload and standalone update notices; the bundled CLI updates with the plugin.\n\nReviews can take longer than two minutes. If the shell tool returns a running\nsession, keep polling that session until it completes rather than starting a\nsecond review.\n\nIf authentication is missing, direct the user to the plugin's `login` skill.\nDo not start browser sign-in as a side effect of this review command.\n\nThe CLI stores the review on the user's Greptile account. The MCP tools\n`list_code_reviews` and `get_code_review` can read it with `source: \"headless\"`;\nMCP authentication is separate from CLI login.\n\nSummarize the findings and offer to fix them. Use `get_code_review` when more\nfinding detail is needed. If the command fails, report the failure and the\nnext action rather than presenting it as a successful review.\n"
}