fcakyon/claude-codex-settings / plugins/codex-advisor/skills/codex-advisor
codex-advisor
VerifiedThis skill should be used when the user asks for a "GPT second opinion", wants a "cross-model review", needs to "check a plan before committing", wants another view after repeated failures, or explicitly invokes "codex-advisor".
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 1.4 KB |
| reviewer-prompt.md | markdown | 1.6 KB |
| scripts/ask_codex.mjs | script | 3.9 KB |
scripts/ask_codex.mjs
View on GitHub#!/usr/bin/env node
// Sends a review request to the Codex CLI and prints only the reviewer's answer.
// The question arrives on standard input. --transcript <file> prepends the recent conversation.
import { readFileSync } from "node:fs";
import { spawnSync } from "node:child_process";
const args = process.argv.slice(2);
const transcriptIndex = args.indexOf("--transcript");
const question = readFileSync(0, "utf8").trim();
if (!question)
throw new Error("ask_codex: pass the review request on standard input");
const clip = (s, n) => (s.length > n ? s.slice(0, n) + " …[truncated]" : s);
const renderBlock = (role, b) => {
switch (b.type) {
case "text":
return `${role}: ${clip(b.text ?? "", 4000)}`;
case "thinking":
return `assistant (thinking): ${clip(b.thinking ?? "", 4000)}`;
case "tool_use":
return `assistant → ${b.name ?? "?"}(${clip(JSON.stringify(b.input ?? {}), 8000)})`;
case "tool_result": {
const c = b.content;
const text =
typeof c === "string"
? c
: Array.isArray(c)
? c.map((x) => x.text ?? "").join(" ")
: JSON.stringify(c ?? "");
return ` ↳ result: ${clip(text, 8000)}`;
}
default:
return "";
}
};
let context = "";
if (transcriptIndex !== -1) {
const transcript = args[transcriptIndex + 1];
const lines = readFileSync(transcript, "utf8").split("\n");
const records = [];
let i = lines.length - 1;
for (; i >= 0 && records.length < 80; i--) {
if (!lines[i]) continue;
let record;
try {
record = JSON.parse(lines[i]);
} catch {
continue;
}
if (record.isSidechain === true || record.isMeta === true) continue;
if (record.type !== "user" && record.type !== "assistant") continue;
const content = record.message?.content;
const rendered =
typeof content === "string"
? `${record.type}: ${clip(content, 4000)}`
: Array.isArray(content)
? content.map((block) => renderBlock(record.type, block)).filter(Boolean).join("\n")
: "";
if (rendered) records.push(rendered);
}
records.reverse();
let recent = records.join("\n\n---\n\n");
const MAX = 160000;
if (recent.length > MAX) recent = "…[older turns truncated for length]\n" + recent.slice(-MAX);
if (i >= 0) recent = `…[this is only the newest ${records.length} turns of a longer session]\n` + recent;
if (records.length)
context = `You are reviewing an in-progress Claude Code session. Below is the recent conversation, most recent last, the same history the built-in advisor tool would see. Weigh the question against this actual context, not just the caller's summary of it.
Long turns are shortened and marked \`…[truncated]\`. Full session: ${transcript}, one JSON record per line, newest last. Grep it for a term, number, or phrase you are chasing; never read it whole.
<recent-conversation>
${recent}
</recent-conversation>`;
}
const prompt = [
readFileSync(
new URL("../reviewer-prompt.md", import.meta.url),
"utf8",
).trim(),
context,
`<question>\n${question}\n</question>`,
]
.filter(Boolean)
.join("\n\n");
const result = spawnSync(
"codex",
[
"exec",
// Pinned so the verdict does not drift with whatever ~/.codex/config.toml happens to set.
"--model",
"gpt-6-astra",
"-c",
'model_reasoning_effort="medium"',
"--sandbox",
"read-only",
"--ephemeral",
"--skip-git-repo-check",
"--color",
"never",
"-",
],
{ input: prompt, encoding: "utf8", stdio: ["pipe", "pipe", "pipe"] },
);
if (result.error) throw result.error;
if (result.signal) throw new Error(`codex exited from signal ${result.signal}`);
// Codex narrates the run on stderr, so it only becomes useful once something failed.
if (result.status !== 0) {
process.stderr.write(result.stderr ?? "");
process.exit(result.status ?? 1);
}
process.stdout.write(result.stdout);
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 466k | Updated | Claude CodeCodex | Verified | Install |
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 466k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 319k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 319k | Updated | Claude CodeCodex | Verified | Install |