oracle-codex by paulrberg/agent-skills
npx skills add https://github.com/paulrberg/agent-skills --skill oracle-codex将 OpenAI Codex CLI 用作只读预言机——仅用于规划、审查和分析。Codex 提供其视角;你负责综合并向用户呈现结果。
沙盒始终为read-only。Codex 绝不允许实施更改。
解析 $ARGUMENTS 以获取:
--reasoning <level> — 覆盖推理强度(low、medium、high、xhigh)。可选;默认根据复杂性自动选择。在任何 Codex 调用之前运行检查脚本:
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
scripts/check-codex.sh
如果它以非零状态退出,则显示错误并停止。对所有 codex exec 调用使用包装器:
scripts/run-codex-exec.sh
| 设置项 | 默认值 | 覆盖方式 |
|---|---|---|
| Model | gpt-5.3-codex | 仅允许列表(参见 references/codex-flags.md) |
| Reasoning | Auto | --reasoning <level> 或用户描述 |
| Sandbox | read-only | 不可覆盖 |
| 复杂性 | 强度 | 超时时间 | 标准 |
|---|---|---|---|
| 简单 | low | 300000ms | <3 个文件,快速问题 |
| 中等 | medium | 300000ms | 3–10 个文件,聚焦分析 |
| 复杂 | high | 600000ms | 多模块,架构性思考 |
| 极限 | xhigh | 600000ms | 完整代码库,关键决策 |
对于可能超过 10 分钟的 xhigh 任务,在 Bash 工具上使用 run_in_background: true 并设置 CODEX_OUTPUT,以便稍后读取输出。
完整的标志文档请参见 references/codex-flags.md。
$ARGUMENTS 解析 query 和 --reasoningscripts/check-codex.sh — 失败则中止根据用户的查询和任何相关上下文(差异、文件内容、先前对话)构建一个聚焦的提示。保持直接——说明你希望 Codex 分析什么以及你需要什么样的输出。不要实施;仅请求分析和建议。
通过包装器使用 HEREDOC 调用。根据上面的推理强度表设置 Bash 工具的超时时间。
EFFORT="<effort>" \
CODEX_OUTPUT="/tmp/codex-${RANDOM}${RANDOM}.txt" \
scripts/run-codex-exec.sh <<'EOF'
[constructed prompt]
EOF
对于 xhigh,考虑在 Bash 工具调用时使用 run_in_background: true,然后在完成后读取 CODEX_OUTPUT。
读取输出文件并注明出处:
## Codex 分析
[Codex 输出 — 如果超过 200 行则进行总结]
---
Model: gpt-5.3-codex | Reasoning: [effort level]
为用户综合关键见解和可操作项。
每周安装量
128
代码仓库
GitHub 星标数
41
首次出现
2026年2月5日
安全审计
安装于
claude-code115
codex110
gemini-cli104
github-copilot104
amp104
kimi-cli104
Use OpenAI Codex CLI as a read-only oracle — planning, review, and analysis only. Codex provides its perspective; you synthesize and present results to the user.
Sandbox is alwaysread-only. Codex must never implement changes.
Parse $ARGUMENTS for:
--reasoning <level> — override reasoning effort (low, medium, high, xhigh). Optional; default is auto-selected based on complexity.Run the check script before any Codex invocation:
scripts/check-codex.sh
If it exits non-zero, display the error and stop. Use the wrapper for all codex exec calls:
scripts/run-codex-exec.sh
| Setting | Default | Override |
|---|---|---|
| Model | gpt-5.3-codex | Allowlist only (see references/codex-flags.md) |
| Reasoning | Auto | --reasoning <level> or user prose |
| Sandbox | read-only | Not overridable |
| Complexity | Effort | Timeout | Criteria |
|---|---|---|---|
| Simple | low | 300000ms | <3 files, quick question |
| Moderate | medium | 300000ms | 3–10 files, focused analysis |
| Complex | high | 600000ms | Multi-module, architectural thinking |
| Maximum | xhigh | 600000ms | Full codebase, critical decisions |
For xhigh tasks that may exceed 10 minutes, use run_in_background: true on the Bash tool and set CODEX_OUTPUT so you can read the output later.
See references/codex-flags.md for full flag documentation.
$ARGUMENTS for query and --reasoningscripts/check-codex.sh — abort on failureBuild a focused prompt from the user's query and any relevant context (diffs, file contents, prior conversation). Keep it direct — state what you want Codex to analyze and what kind of output you need. Do not implement; request analysis and recommendations only.
Invoke via the wrapper with HEREDOC. Set the Bash tool timeout per the reasoning effort table above.
EFFORT="<effort>" \
CODEX_OUTPUT="/tmp/codex-${RANDOM}${RANDOM}.txt" \
scripts/run-codex-exec.sh <<'EOF'
[constructed prompt]
EOF
For xhigh, consider run_in_background: true on the Bash tool call, then read CODEX_OUTPUT when done.
Read the output file and present with attribution:
## Codex Analysis
[Codex output — summarize if >200 lines]
---
Model: gpt-5.3-codex | Reasoning: [effort level]
Synthesize key insights and actionable items for the user.
Weekly Installs
128
Repository
GitHub Stars
41
First Seen
Feb 5, 2026
Security Audits
Gen Agent Trust HubPassSocketPassSnykPass
Installed on
claude-code115
codex110
gemini-cli104
github-copilot104
amp104
kimi-cli104
AI Elements:基于shadcn/ui的AI原生应用组件库,快速构建对话界面
67,500 周安装