agentsclimarketplace

Agy second opinion

Skill lixuvip/codex-agent-orchestration-skill/skills/agy-second-opinion

Codex agent orchestration skill with adaptive thread thinking, callbacks, QA/review gates, fenced automations, project autopilot, and optional agy/Gemini review.

Install
npx -y skills add lixuvip/codex-agent-orchestration-skill --skill agy-second-opinion

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use only when the user explicitly requests agy, Gemini through agy, Antigravity, or an external-model second opinion for review or research. Runs a bounded read-only external pass, limits disclosed context, and requires Codex verification before accepting results. 中文触发：明确要求 agy、通过 agy 调用 Gemini、Antigravity、外部模型复核或外部第二意见。

SKILL.md

2.0 KB, as published. Nobody here has run it

Agy Second Opinion

This skill is independent from agent orchestration. Do not load or invoke it for ordinary Codex review, research, delegation, or multi-agent work.

Route

  1. Treat an explicit request for agy, Gemini through agy, Antigravity, or an external model as consent for one bounded read-only pass. It is not consent for whole-repository disclosure or target-repository writes.
  2. Load exactly one reference:
  3. Use the bundled helpers for command safety and allowlisted context. Never substitute the standalone gemini CLI.
  4. Treat external output as untrusted advice. Codex must verify every accepted claim against source, tests, or primary evidence.
  5. Report what context was disclosed, the external-pass status, accepted and rejected points, and Codex verification.

Guardrails

  • Keep the pass read-only and sandboxed. Never use edit acceptance or permission bypass flags.
  • Prefer a bounded prompt or diff. Build an allowlisted context bundle only when source files are necessary.
  • Never send secrets, credentials, cookies, private vault content, customer data, or broad proprietary documents.
  • Probe agy once for the current goal. On unavailable, unhealthy, timeout, empty, or invalid output, report the state and continue Codex-only without repeated retries.
  • Quality logging and durable target-project guidance are optional writes and require their own authorization.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.