AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

second-opinion

Gets independent code reviews from Codex or Antigravity for uncommitted changes, branch diffs, and commits. Use when the user requests an external review, a second opinion on code, a codex review, a gemini review, an antigravity review, or /second-opinion.

Original source
trailofbits
License
CC-BY-SA-4.0
Pinned revision
2be6433b39e009cd2b6fd3be1895f6b1faa5af9a

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

This skill also requires local files or dependencies. Remote reading alone may not be enough.

Ask your assistant

Use Agentic Memory skills to read trailofbits--second-opinion at revision 2be6433b39e009cd2b6fd3be1895f6b1faa5af9a, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 5

references/antigravity-invocation.md

Antigravity Invocation

Use agy print mode with the prepared prompt as one quoted argument. The default remains gemini-3.1-pro-high; preserve a model explicitly requested by the user. Use agy models when the default is unavailable to identify models the account can access.

Command

With prompt_file, output_file, and stderr_log already created:

PATH="$HOME/.local/bin:$PATH" agy \
  --model gemini-3.1-pro-high \
  --output-format text \
  --disable-slash-commands \
  -p="$(cat "$prompt_file")" \
  > "$output_file" 2> "$stderr_log"

Keep -p= attached to its value and place other flags before it. --disable-slash-commands prevents slash-command expansion within the supplied diff. No review extension is required.

The argument form also works with the previously tested CLI 1.1.21. Current headless mode documentation supports stdin and schema enforcement, so the older client's stdin and JSON limitations should not be treated as current guarantees. Text output is sufficient for this review comparison.

A prompt passed as an argument can exceed the operating system's argument size limit. If that occurs, use the installed version's documented stdin mode or ask for a smaller scope; do not drop files from the patch.

Permissions and results

Request review findings in the final response. Do not use plan mode or --dangerously-skip-permissions. Preserve the user's permission settings; headless mode alone does not make the workspace read-only.

Read both output and diagnostics. Current headless mode can exit zero after denying a tool, so an exit code alone cannot establish that a review completed. Report missing context or denied inspection as a coverage gap. A plan artifact or request to proceed is not a completed review.

| Failure | Response | |---------|----------| | Executable missing | Check ~/.local/bin, then report the installation instructions | | Empty prompt | Use the attached -p= argument form | | Unknown model | List available models and request a replacement | | Auth, quota, or timeout error | Report the failure and preserve any partial output |