القائمة
Query OpenAI Codex (GPT-5.2-codex) for second opinions, alternative implementations, or specialized knowledge. Use when you want to consult another AI model, need a different perspective on a coding problem, or want to verify an approach.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Verify GSD setup - checks Claude, Codex, skills, and MCP servers are working correctly.
Run entire milestone autonomously with Codex validation. Auto-detects current milestone, executes all phases, and runs audit.
Get a second opinion from OpenAI Codex (GPT-5.2-codex) on code, architecture, or implementation decisions. Use when you want independent verification or an alternative perspective.
Autonomous phase execution with Codex validation. Runs plan/execute/verify cycles in a bash loop until done or halted.
| name | codex-oracle |
| description | Query OpenAI Codex (GPT-5.2-codex) for second opinions, alternative implementations, or specialized knowledge. Use when you want to consult another AI model, need a different perspective on a coding problem, or want to verify an approach. |
| argument-hint | [prompt] |
Invoke OpenAI's Codex CLI to get responses from GPT-5.2-codex.
codex CLI installed (npm i -g @openai/codex)codex auth login (uses account auth, not API key)Run the script with a prompt. Codex runs in read-only mode and sees the same working directory.
# Determine script location (project-local or global install)
CODEX_SCRIPT=".claude/skills/codex-oracle/scripts/ask_codex.sh"
[[ ! -f "$CODEX_SCRIPT" ]] && CODEX_SCRIPT="$HOME/.claude/skills/codex-oracle/scripts/ask_codex.sh"
# Default: gpt-5.2-codex with medium reasoning
bash "$CODEX_SCRIPT" "What's the most efficient way to implement a bloom filter in Rust?"
# With model override
bash "$CODEX_SCRIPT" "Review error handling in this file" codex-mini-latest
# Full output (all thinking/exec traces)
bash "$CODEX_SCRIPT" "Debug this complex race condition" gpt-5.2-codex xhigh 600 full
| Arg | Default | Description |
|---|---|---|
$1 | (required) | The prompt |
$2 | gpt-5.2-codex | Model (gpt-5.2-codex, codex-mini-latest) |
$3 | xhigh | Reasoning effort (low, medium, high, xhigh) |
$4 | 600 | Timeout in seconds (10 min) |
$5 | brief | Output verbosity (brief = final answer only, full = all traces) |
xhigh effort for complex multi-step reasoning